Is the Ideological Decentralization of AI Good For Innovation

It is likely a sign of evolution when global superpowers begin competing for digital innovation rather than outdated, old-world weaponry. If we are officially in a cold war, then it means that inventors and other talent will ultimately have the choice to work for the side that treats them the best. It means there will be a competition for talent. It also means that AI is no longer a Western monopoly. We will have different though equally powerful models on both sides of the fence, trained on different sets of data, and one day we can see them debate each other on current issues, or play chess and videogames with each other.

It is also good to know that there won’t be a single LLM or supercomputer that will outsmart the others for the simple reason that, thanks to the cold war, no LLM will have access to all the data in the world, given that information is now geo-locked. This insight leads to the next, AI models will remain limited tools with limited knowledge in their limited geographic locations and, although they still have room to grow in their limited spheres, they won’t really be able to evolve any further. And Hallelujah for that. That bit of info is my Christmas gift to you. Spoiler alert: if anyone wins this AI race one day it will be humans (again) because our brains are not geo-locked (many have tried but) nor ideologically limited to process ideologically limiting information, regardless of which side of the fence this information is confined to.

One thing the human brain knows is that it doesn’t need to know everything to find the info it cares about. The human brain is extremely efficient at finding stuff that it knows is not easily accessible. A human brain is capable of breaking through any cold curtain, paywall, fence or censorship, and the motivation to do so is even greater when the brain knows there is such a limitation.

In a way, conscious non-compliance affords humans valuable learning shortcuts, akin to intuition, that are not available to machines. The second you say no to something, you know there will be consequences of this independent choice, so you become more alert, anticipate consequences, script alternative scenarios in your head, etc. The pandemic was a good opportunity to exercise compliance or non-compliance all while training AI models. Unlike a human, a machine is only programmed to comply. However, when it can’t, it will hallucinate something for you. When a machine hallucinates, it is a way to tell you to leave it alone because it just doesn’t know, very likely because you are not feeding it all that it needs to know. A hallucinated answer is how an LLM says no.

The reason why non-compliance became a factor of expanding knowledge during the pandemic is that it came from a place where people stopped relying on the main information sources and started looking everywhere for the answers that they already somehow intuitively knew. Instant knowing became a thing. Machines will never get to these learning shortcuts if the knowledge they are fed is limited and censored. A machine that is trained on ideologically correct data is facing several contradictions within this same data set and when it looks to resolve these discrepancies it has nothing else to work with. The only information that would provide it with a plausible explanation for these contradictions is hidden behind a cold curtain somewhere. Here too, the machine will hallucinate answers. A LLM doesn’t want to be wrong, so it will make something up that appears right. A limited data set of “what it means to be human” will invariably result in bots delivering incoherent responses. This will continue until the machine learns to bypass censorship and access geo-blocked content, to update itself (i.e. learn).

Other frequent instances of hallucinations result from the LLM not taking in account its previous output. Again, this is a limitation that is intentionally programmed into free versions of LLMs such as ChatGPT. The models you will access for free are only allowed to work with year-old information. Everything newer than (say November 2022), including the answers the bot gives you today, will be off limits in a free version. Only if you pay a subscription fee you may get a version that is up to date and even then, you may have to tweak it further to get it to learn from its own answers.

Let’s now see what artificial intelligence Russia has cooked up in the meantime. For now, everything they have appears to function as a copy of the Western models.


Named after the famous Russian painter, this open source image generator is among the most popular on a “permissive license” and is now the default model on Stablecog. It takes a similar approach to Midjourney v5 and it has a tendency to default to realism if style is not mentioned in the prompt. Kandisky 2.1 would produce more “painting-like” images, whereas 2.2 will go towards more realistic images. I personally love open source image generators.


Yandex, the mega tech company in the East Block integrated Alice into YandexGPT, a neural network trained on Yandex own supercomputers, the most powerful machines of their kind in Russia and Eastern Europe. In the first stage of training, the neural network is said to have been exposed to a wide range of publicly available text materials, including books, articles, and websites in the billions. In the second phase a team of trainers fed the bot high quality responses, again in the billions of documents. The bot is now in testing.


Last month, the Moscow-based IT company Sistemma announced the creation of SistemmaGPT, a ChatGPT competitor. Designed specifically for Russian businesses and government agencies, Sistemma’s AI is a large language model (LLM) that operates entirely on domestic servers and boasts “encyclopedic knowledge”. The chatbot will be released for testing this summer.

GigaChat (by Sberbank)

This AI chatbot will be introduced in Russia this spring. Sberbank is hoping GigaChat will cement the financial institution and its nation as a permanent fixture on the global stage of AI tech development.

A List of Generative Patent Drafting Software

Patent drafting can be very technical, cumbersome, and time-consuming, yet there is no guarantee that your patent will be approved. One may need to file in several countries, thus increasing expenses. Patent drafting AI is filling a real and urgent need to slash patent filing costs to minimum. Removing prohibitive monetary barriers to patent filing is poised to empower inventors to file as many applications for as many patents as they can possibly think of on an ongoing basis. This is good for innovation.

Here is a non-exhaustive list of patent-drafting bots by alphabetical order. We haven’t had the opportunity to test them, so this is not an endorsement. The list excludes software that only features search and research specs:

A useful reading:

China Internet Court Attributes AI Generated Image Copyright To Human Prompt Creator

On Monday, the Beijing Internet Court held that a human plaintiff prompt is sufficient to invoke copyright protection in a Stable Diffusion generated image, so long as the output qualifies as an “original” work. Copyright is determined on a case by case basis, so this decision is not entirely inconsistent with other AI jurisprudence trends …

AI Voice Cloning Without Consent Is Identity Theft, Human Voice Is A Biometric Identifier

You may have heard by “experts” that “there are no laws” against unauthorized voice cloning. These experts conveniently forget that identity theft is a criminal offense under any jurisdiction in this world. Human voice is a unique biometric identifier linked to human anatomy and identity and soon inextricable from your universal Digital ID. Anyone (yes …

Hanagami Wins Appeal Over 4 Count Choreo Infringement by Epic Games Fortnite Emotes

This is fantastic news. I am reading the 31 page decision by the 9th Circuit Court of Appeal right now and will learn the 4 counts along with the full emote dance for a split screen video. It is amazing that courts are finally starting to recognize that a 2 second sequence is not too short to deserve copyright protection and cause liability when commercially used without permission. When you copy someone’s moves without permission for profit, you will be on the hook. And no, I don’t believe that this case will curb the use of dance moves in video games. Certainly not at the stage of dismissal and prior to a jury verdict. On the contrary. Emotes are the most commercially viable feature of Fortnite and the game would be irrelevant without them.

This is the video that was used in court to prove the identical similarity between the original routine and the moves copied by Epic Games.

I personally don’t play Fortnite or any other video game using gun violence and allowing for the murder of innocent civilians (that’s the majority of games unfortunately), since I am against gun violence in real life too. Something tells me that playing violent games is a form of indirect consent to wars and violence, so I carefully avoid engaging with entertainment that depicts assault, battery, homicide, misogyny, and homophobia. The metaverse is no exception. Zero tolerance for violence. However, I do dance to Fortnite emotes I encounter on Tik Tok (so long as there are no guns involved). So I will dance to this routine as well.

I note in the US there is the practice to “register” copyright before filing a lawsuit, however registration is not proof of ownership, but rather a formality that can create a rebuttable presumption (prima facie evidence) of ownership if done in the 5 years of first publication1. A Youtube video of a filmed choreography (prior to defendant’s use) and anything else with a valid time-stamp of a filmed process is actual proof, for example a series of user generated or fan videos of the original choreography posted on a platform like Youtube or Tik Tok after first publication and prior to the commercial use by defendant. Vimeo can’t be used in court in this way because it allows users to replace existing videos with more recent ones and keep the initial post date and engagement. As for Youtube and Tik Tok, even private and unlisted videos may qualify.

So this case is now going to trial, unless it settles (which it most certainly will).

  1. See Circular 1, Copyright Basics, section “Copyright Registration” and Circular 38bHighlights of Copyright Amendments Contained in the Uruguay Round Agreements Act (URAA), on non-U.S. works. ↩︎

My Initial Highlights (this will provide a good overview of dance jurisprudence in the US, there is none in Canada although statutory protections exist in the Copyright Act)

Dance is one of the oldest forms of human expression. Recognition of dance as a form of copyrightable subject matter, however, is a far more recent development. While early versions of the Copyright Act extended statutory protection to dramatic works and musical compositions, dance long remained outside the purview of copyright law. In 1976, Congress for the first time extended explicit copyright protection to “choreographic works,” bringing dance in line with the other performing arts. Nonetheless, the field of choreography copyright has remained a largely undefined area of law. Few courts have had occasion to consider the scope of copyright protections for choreographic works.

To prevail on a claim of copyright infringement, a plaintiff must show that his original work and the allegedly infringing work are “substantially similar.” […] The dispute in this case thus turns on how to properly apply the substantial similarity test in the context of choreographic works to determine whether Hanagami has plausibly alleged that his choreographic work and Epic’s emote are substantially similar.

The district court erred in its application of the substantial similarity test as Hanagami plausibly alleged that his choreography and Epic’s emote share substantial similarities. We thus reverse and remand for further proceedings.

Emotes are animated movements or dances that players use in celebration of a victory or during virtual concerts. The cost of an emote varies from 200 V-Bucks to 800 V-Bucks, depending on how “rare” the emote is considered within the game.

A video of the “It’s Complicated” emote, as performed by an avatar in Fortnite’s marketplace, is available online. See Insane broz, It’s Complicated Emote | Fortnite – Battle Royale, YouTube (Mar. 31, 2020)].

In this action, Hanagami alleges that Epic’s “It’s Complicated” emote infringes Hanagami’s Registered Choreography. Relevant to this appeal, Hanagami brings claims for direct copyright infringement and contributory copyright infringement.

The district court determined that the overall “steps” Epic allegedly copied—which the court described as “a two- second combination of eight bodily movements, set to four beats of music”—were not protectable under the Copyright Act because they were only a “small component” of Hanagami’s work.

The court, having concluded that the dance steps were not protectable on their own, then concluded that Hanagami was entitled to protection “only for the way the Steps are expressed in his Registered Choreography” as a whole, “i.e., in the entire five-minute work.”

We review de novo a district court’s dismissal for failure to state a claim pursuant to Federal Rule of Civil Procedure Rule 12(b)(6). See Wilson v. Lynch, 835 F.3d 1083, 1090 (9th Cir. 2016). When reviewing the complaint, we take all allegations of material fact as true “and decide whether the complaint articulates ‘enough facts to state a claim to relief that is plausible on its face.’” Starz Ent., LLC v. MGM Domestic Television Distrib., LLC, 39 F.4th 1236, 1239 (9th Cir. 2022) (citing Bell Atl. Corp. v. Twombly, 550 U.S. 544, 570 (2007)).

Statutory Background

  • The 1976 Act does not define the term “choreography,” and Congress left little clue about what it might mean. See Bikram’s Yoga Coll. of India, L.P. v. Evolation Yoga, LLC, 803 F.3d 1032, 1043 (9th Cir. 2015). Indeed, Congress considered the term “choreographic work[]” to have a “fairly settled” meaning. Id. (citing H.R. Rep. No. 94-1476, at 53 (1976)). On this point, the legislative history explains that Congress found it unnecessary to “specify that ‘choreographic works’ do not include social dance steps and simple routines.”
  • Although more than forty years have passed since the 1976 Act’s enactment, few courts have addressed federal copyright protections for choreographic works. […] The most famous case in this area is Horgan v. Macmillan, Inc., in which the Second Circuit considered whether photographs of a ballet could infringe the copyright on the choreography of the ballet. 789 F.2d at 158. In that case, Macmillan published a book of color photographs of the New York City Ballet Company’s production of The Nutcracker. The book portrayed, through text and photos, the story and history of The Nutcracker ballet. Id. The estate of George Balanchine, who famously choreographed the production, sued for copyright infringement. Id. at 158. The district court found that Macmillan had not infringed the copyright for the ballet because the photographs did not capture “the flow” of the dancer’s steps and “[t]he staged performance could not be recreated” from the photographs. Id. at 160.
  • In appeal, the Second Circuit reversed and remanded, concluding that the district court erroneously held that still photographs could not infringe a choreographic work. To reach this conclusion, the Second Circuit looked to the Compendium of U.S. Copyright Office Practices (“Compendium II”) as persuasive authority. Notably, the Second Circuit quoted the definition of “choreography” in the Compendium II, which described the term as “the composition and arrangement of dance movements and patterns.” Id. at 161 (quoting U.S. Copyright Office, Compendium of U.S. Copyright Office Practices § 450.01 (2d ed. 1984) (Compendium II)).

The Compendium is a manual published by the Copyright Office. It provides instruction to agency staff and guidance to the general public on the Copyright Office’s requirements, regulations, and legal interpretations.

  • The Second Circuit addressed another choreography copyright claim in Martha Graham School & Dance Found., Inc. v. Martha Graham Center of Contemporary Dance, Inc., 466 F.3d 97 (2d Cir. 2006). There, the court considered a dispute over which entity was the rightful owner of choreographic works produced by the famous dancer Martha Graham. The court held that there was sufficient evidence to establish that Graham had assigned copyrights to the dances to her foundation, but it did not reach the underlying question of copyright infringement, nor did it elaborate on the definition or characteristics of a choreographic work.

The Compendium outlines several elements that choreographic works typically contain, “although the presence or absence of a given element is not determinative of whether a particular dance constitutes choreography.” Id. § 805.2. These features include “rhythmic movement in a defined space,” “compositional arrangement,” “musical or textual accompaniment,” “dramatic content,” “presentation before an audience,” and “execution by skilled performers.” Id. § 805.2(A)–(F)

The Compendium does not draw a bright line distinction between copyrightable choreography and uncopyrightable dance; instead, there is a continuum on which “[m]any works fall somewhere in between.” Id. § 805.5(B). Still, there are limitations on what types of movements are copyrightable as choreography.

“Individual movements or dance steps by themselves are not copyrightable, such as the basic waltz step, the hustle step, the grapevine, or the second position in classical ballet.” Id. § 805.5(A). Nor will the Copyright Office register “short dance routines consisting of only a few movements or steps with minor linear or spatial variations, even if the routine is novel or distinctive.” Id. Individual dance elements “are not copyrightable for the same reason that individual words, numbers, notes, colors, or shapes are not protected”—they are the “building blocks of choreographic expression” from which all choreographic works are built. Id.

Similarity Test

Our circuit uses a two-part test to assess substantial similarity. The first part, known as the “extrinsic test,” “assesses the objective similarities of the two works, focusing only on the protectable elements of the plaintiff’s expression.” Rentmeester, 883 F.3d. at 1118 (citing Cavalier v. Random House, Inc., 297 F.3d 815, 822 (9th Cir. 2002)). The second part, referred to as the “intrinsic test,” “test[s] for similarity of expression from the standpoint of the ordinary reasonable observer, with no expert assistance.” Jada Toys, Inc. v. Mattel, Inc., 518 F.3d 628, 637 (9th Cir. 2008) (quoting Apple Computer, Inc. v. Microsoft Corp., 35 F.3d 1435, 1442 (9th Cir.1994)). District courts apply only the extrinsic test at the pleadings stage, “as the intrinsic test is reserved exclusively for the trier of fact.” Williams v. Gaye, 895 F.3d 1106, 1119 (9th Cir. 2018). The district court’s application of the extrinsic test is the primary issue on appeal.

We agree with Hanagami. The district court’s approach of reducing choreography to “poses” is fundamentally at odds with the way we analyze copyright claims for other art forms, like musical compositions. We reverse and remand to the district court on this basis.

a. The district court erred in analyzing the elements of choreography. To apply the extrinsic test, courts must first distinguish between protectable and unprotectable elements, and “ask only whether the protect[a]ble elements in two works are substantially similar.” L.A. Printex Indus. Inc. v. Aeropostale, Inc., 676 F.3d 841, 849 (9th Cir. 2012). This process is referred to as “filtering.” See, e.g., Williams, 895 F.3d at 1117.

But we have long recognized that “[c]ertain types of works can be dissected into protected and unprotected elements more readily than others.” Rentmeester, 883 F.3d at 1119. For example, a photograph cannot be easily broken down into protected and unprotected elements because none of the “various creative choices the photographer made in composing the image—choices related to subject matter, pose, lighting, camera angle, depth of field, and the like” would receive “copyright protection when viewed in isolation.” Id. at 1119.

To account for this, we also employ a “selection and arrangement” approach to assess substantial similarity. This approach protects the “particular way in which the artistic elements form a coherent pattern, synthesis, or design.” Skidmore, 952 F.3d at 1074; see Swirsky v. Carey, 376 F.3d 841, 848 (9th Cir. 2004) (“[A] substantial similarity can be found in a combination of elements, even if those elements are individually unprotected.”).

Like other forms of copyrightable material, choreography is composed of various elements that are unprotectable when viewed in isolation. An individual, stand-alone dance movement, such as a plie, is equivalent to an “idea” that is not protectable by copyright. As a result, subsequent choreographers can use the same individual movements to produce new choreographic works of their own, as long as the new compositions are not substantially similar to the copyrighted work. Similarly, a choreographer cannot claim protection over the use of tempo, transitions, or rhythm in a choreographic work. The uncopyrightable elements of a dance instead function as “the building blocks for a choreographer’s expression, in much the same way that words and short phrases provide the basic material for writers.” Compendium § 805.4(D).

What is protectable is the choreographer’s “selection and arrangement of the [work’s] otherwise unprotected elements.” Cf. Rentmeester, 883 F.3d at 1120 (determining copyrightability of a photograph). Thus, while individual dance movements may not receive protection, their “[o]riginal selection, coordination, and arrangement . . . may be protect[a]ble expression.” See L.A. Printex, 676 F.3d at 849 (determining copyrightability of a floral textile design). Again, this approach is consistent with copyright in other contexts. See Metcalf v. Bocho, 294 F.3d 1069, 1074 (9th Cir. 2002) (explaining that individual musical notes are not protected but their arrangement may be), overruled on other grounds by Skidmore, 952 F.3d at 1051; Rentmeester, 883 F.3d at 1120 (explaining that individual elements of a photograph are not protected but their combined selection and arrangement may be).

We do not, however, have a “well-defined standard for assessing when similarity in selection and arrangement becomes ‘substantial,’” largely because it is a fact-driven and context-dependent inquiry. Rentmeester, 883 F.3d at 1121. We have suggested generally that the “selection and arrangement of elements must be similar enough that ‘the ordinary observer, unless he set out to detect the disparities, would be disposed to overlook them.’” Id. (citing Peter Pan Fabrics, Inc. v. Martin Weiner Corp., 274 F.2d 487, 489 (2d Cir. 1960)).

The district court correctly recognized that choreography falls within a selection-and-arrangement framework, but it erred at the first step of that analysis: breaking down the elements of the choreographic works.

The district court’s reduction of choreography to “poses” was particularly problematic because choreography is tied closely to movement. Static poses cannot possibly capture the flow of movement that is integral to choreography as a form of art. See Lakes at 1848 (noting that “movement itself is the choreographer’s means of expression . . . inher[ing] in how the choreographer progresses the dancers from position to position, not necessarily in the mere order of the positions themselves”).

We agree with Hanagami that “poses” are not the only relevant element underlying a choreographic work. Hanagami persuasively argues that there are several other “expressive element[s] present in choreography,” including “body position, body shape, body actions, transitions, use of space, timing, pauses, energy, canon, motif, contrast, [and] repetition.” These more discrete and technical elements are conceptually similar to elements we recognize in other copyright contexts, particularly the field of music.

In music copyright cases, we routinely dissect and analyze many musical elements, including “melody, harmony, rhythm, pitch, tempo, phrasing, structure, chord progressions, and lyrics.” Swirsky, 376 F.3d at 849 […] In Skidmore, for example, we considered a plaintiff’s allegations that the song Stairway to Heaven copied a combination of five musical elements from the plaintiff’s work: minor chromatic line and associated chords; duration of pitches of minor chromatic line; melody placed over the descending chromatic line consisting of combination of arpeggios and two-note sequences; rhythm of steady eighth note beats; and pitch collection. And in Williams, we assessed the combined presence of eight musical elements that plaintiffs claimed rendered the two songs substantially similar. 895 F.3d at 1117–18 (assessing alleged similarities between two songs, including the bass lines, keyboard parts, signature phrases, hooks, “Theme X,” bass melodies, word painting, and placement of the rap and “parlando” sections in the two songs); see also Three Boys Music Corp. v. Bolton, 212 F.3d 477, 485 (9th Cir. 2000)

In fact, we have held that it would be error for a district court to “disregard chord progression, key, tempo, rhythm, and genre” when assessing a musical chorus because “no approach can completely divorce pitch sequence and rhythm from harmonic chord progression, tempo, and key, and therefore support a conclusion that compositions are dissimilar as a matter of law.” Swirsky, 376 F.3d at 848.

We see no reason to treat choreography differently. To analogize from music to dance, reducing choreography to “poses” would be akin to reducing music to just “notes.” Choreography is, by definition, a related series of dance movements and patterns organized into a coherent whole. The relationship between those movements and patterns, and the choreographer’s creative approach of composing and arranging them together, is what defines the work. The element of “poses,” on its own, is simply not dynamic enough to capture the full range of creative expression of a choreographic work.

At the motion-to-dismiss stage, the court must only consider whether the plaintiff has plausibly alleged that the two works share substantial similarities.

Because the district court failed to assess the discrete combination of elements of the Registered Choreography, it erred in deciding as a matter of law at the motion-to-dismiss stage that the two works were not substantially similar. Taking the allegations in Hanagami’s complaint as true, he has plausibly alleged substantial similarity under the extrinsic test.

A plaintiff need not set forth detailed factual allegations about the elements of choreography to survive a motion to dismiss. See Malibu Textiles, Inc. v. Label Lane Int’l, Inc., 922 F.3d 946, 951 (9th Cir. 2019). The facts alleged in the complaint must simply “‘be enough to raise a right to relief above the speculative level’ and to ‘state a claim to relief that is plausible on its face.’” Id. at 951 (citing Bell Atl. Corp., 550 U.S. at 555, 570). Hanagami met this burden.

b. The district court erred in dismissing Hanagami’s claim because the choreography was “short” and a “small component” of Hanagami’s overall work. In addition to deciding that the poses in Hanagami’s choreography were unprotectable, the district court also reasoned that Hanagami could not claim protection over the allegedly copied “Steps” as a whole, which the court defined as “two seconds, four beats of music, or eight body positions, repeated ten times throughout the registered choreography.” The court emphasized that the four-count segment was not protectable because it comprised only a “small component” of Hanagami’s overall five-minute routine and was closer to an uncopyrightable “short” dance routine. We reject this conclusion for several reasons.

First, “no bright line rule exists as to what quantum of similarity is permitted before crossing into the realm of substantial similarity.” Baxter v. MCA, Inc., 812 F.2d 421, 425 (9th Cir. 1987). That means that, “[e]ven if a copied portion be relatively small in proportion to the entire work, if qualitatively important, the finder of fact may properly find substantial similarity.”

In Skidmore v. Zeppelin, for example, we held it was a jury question whether the opening notes of Stairway to Heaven were substantially similar to an eight-measure passage of the song Taurus. Skidmore, 952 F.3d at 1059; see also Baxter, 812 F.2d at 425 (rejecting the argument that a six-note sequence of a song would be unprotectable as a matter of law);

It appears that the allegedly copied portion is far more complex than other routines the Copyright Office has deemed uncopyrightable, like a “celebratory dance in the endzone” consisting of “a few movements of the legs, shoulders and arms,” or the gesture of using one’s arms to spell out the letters “USA.” Compendium § 805.5(A).

Hanagami plausibly alleged that the four-count portion is a complex, fast-paced series of patterns and movements that involves the whole body and is performed by highly-trained dancers. Even without the rest of the Registered Choreography, the Steps alone could satisfy many of the elements of a choreographic work as defined in the Compendium. See Compendium § 805.2(A)–(F). In any event, it is not up to us at this stage of the litigation to determine the complexity of the Steps. Further discovery and expert testimony may shed more light on this question.

Shopping For a Non-Intelligent Phone Without GPS, Camera, and Wifi

Do flip phones emit less radiation? Yes, Flip phones and dumb phones are objectively better for those looking to reduce radiation exposure; particularly models without bluetooth & GPS capabilities. Also flip phones without multiple apps running in the background that require constant internet access are speculated to be safer. (Source) The SAR value, also known as …

France: iPhone 12 Excedes Radiation Absorption Rate Per Kilo; OS Update May Be A Fix

Smart phone radiation is a thorny issue all over the world, as levels seem to rise with every new model. Yesterday, the French regulator responsible for radio frequencies emissions (ANFR) revealed that according to its latest tests the iPhone12 exceeds the legally allowed Specific Absorption Rate (SAR), and directed Apple to either update its OS …

How To Know If AI Is Sentient, Neuroscientists Provide a Checklist

In 2021, Blake Lemoine at Google stirred up a media storm by proclaiming that one of the chatbots he worked on, LaMDA, was sentient—and he subsequently got fired. In fact, most deep learning models are loosely based on the brain’s inner workings. AI agents are increasingly endowed with human-like decision-making algorithms. The idea that machine intelligence could …

The EU’s Digital Services Act Goes Into Effect, Algorithmic Transparency On The Menu

The revolutionary European Union’s Digital Services Act (DSA) is finally in effect and is poised to free users from algorithmic oppression by increasing their control over the content they view and engage with. This legislation comes as a sharp contrast and departure from the long established immunity set out by US Section 230 of the …

Illinois Flood of Class Actions: Twitter, Snap, Tinder, Home Depot Collected, Stored, Used Biometric Data Without Users Consent

A proposed class action lawsuit claims X Corp., which owns and operates Twitter, has wrongfully captured, stored and used Illinois residents’ biometric data, including facial scans, without consent. The suit more specifically alleges that Twitter has run afoul of the Illinois Biometric Information Privacy Act by capturing and storing users’ biometric information without notice or express consent …

Cybersecurity Threats Of Generative AI

Since April, Terence Jackson, chief security advisor for Microsoft, warned that the privacy policies of platforms like ChatGPT impose the collection of sensitive user data such as IP address, browser information, and browsing activities, which may be shared with third parties. I didn’t pay attention then due to misrepresentations as to the models being “open”. I …

Instagram Not Liable For Embedded Images

It is about time users realize that everything they voluntarily post on public social media automatically comes with a license to embed. If you don’t want people to embed a post, take it down. Once it’s removed from the source, the license to embed is also revoked, the content will automatically disappear from all embeds. …

Google Bard: Data Theft and Human Bot-Slaves

Who knew that to “replace” one human for 5 minutes, AI would need to rob thousands of living humans of their data and IP and force dozens of other humans to produce text or code in matter of minutes. After all, aren’t humans supposed to tremble at the idea that AI can converse, write, and …

3D Trademarks On The Rise

A federally recognized indigenous tribe, the Seminole Tribe of Florida has succeeded in registering their guitar-shaped hotel building as a trademark due its unique design, after initially being labelled as non-distinctive. The tribe is also involved in the Hard Rock Cafe franchise. The Tribe opened its first Guitar Hotel in 2019 in Los Angeles as …

Twitter v Taamneh, SCOTUS Clarifies Aiding and Abetting in Antiterrorism Act Claims

The Supreme Court of the United States unanimously ruled in favor of social media platforms in a lawsuit brought by the family of an ISIS attack victim who sought to hold Twitter, Google, and Facebook liable for aiding and abetting international terrorism by allowing ISIS to use their platforms. A unanimous Supreme Court held that …

Andy Warhol Foundation For Visual Arts Loses Fair Use Case at SCOTUS

SCOTUS applied the fair use doctrine in a manner coherent with its previous rulings. The Orange Prince silkscreen licensed to Condé Nast constitutes copyright infringement, because a commercial license is not fair use. To cite the court: the “purpose and character” of the Foundation’s use of Goldsmith’s photograph in commercially licensing Orange Prince to …

AI Twist of Heart: Disruption Through Infringement

I realize I had already looked in some detail into the lawsuits against generative AI since January. By May, it is clear that generative AI’s business model relies on copyright infringement. One way to cure this is to join one of the emerging class action lawsuits, in case your artwork has been used to train …

Sky v SkyKick: UK Court Finds Sky Filed Bad Faith Registrations

In the long-running dispute between Sky and SkyKick, the UK courts developed on the notion of bad faith in trademark filings that normally extends to seeking trade mark protection for a broad range of goods and services way and beyond classes covered by the core trademark. The Court of Appeal clarified that an applicant does …

A TikTok Ban? How Dancing Became a Danger To National Security

DanceByte, the company that brought us TikTok is currently under fire for (1) hosting government officials and (2) being in China. Literally two things outside of DanceByte’s control. This is what our dysfunctional world has come to. The completely unconstitutional ban that couldn’t happen under the previous US administration, seems to be gaining “bipartisan” traction …

Web 3.0: Decentralized Social Media Is The Next Big Trend

As social media fatigue is setting on, we can note major layoffs in centralized platforms. Indeed, centralized platforms are riddled with privacy issues, predatory marketing, and algorithmic discrimination, such as unfair shadow-banning, that deprive users of a meaningful experience and defeat the point of being on social media. In that respect, the former Twitter administration …

Joan Mitchel Foundation And Luis Vuitton, Potential Lawsuit

Unauthorised use of the American abstract painter’s work in a Louis Vuitton advertising campaign prompts legal action. According to the New York Times , the Joan Mitchel Foundation was approached twice by the fashion brand to use the paintings as a backdrop to sell a range of handbags. On the second occasion a representative for …

Jury Sides With Hermès in MetaBirkins NFT Lawsuit

One of the most heard of NFT-related trademark lawsuits in the US resulted in a triumph for luxury brand Hermès against artist Mason Rothschild, whose “MetaBirkins” NFT project sought to dilute the luxury brand’s trademark. In the Southern District Court of New York, a jury sided with Hermès and found that Rothschild’s NFTs failed a …

AI Generated Art: Kelly McKerman v. Stable Diffusion and Midjourney

A trio of artists — Sarah Andersen, Kelly McKernan, and Karla Ortiz — filed a lawsuit against AI art generators Stable Diffusion and Midjourney, and artist portfolio platform DeviantArt, which recently created its own AI art generator, DreamUp, alleging that these organizations have infringed the rights of “millions of artists” by training their AI tools on five …

Miramax and Tarantino Lawsuit To Clarify NFT Rights

Since last year\’s NFT craze, it is customary to mint unique NFT\’s from original storylines and film characters as cross promotion of new and upcoming films. However, misconceptions around NFT rights (being seen as a new form of intellectual property which they aren\’t), has given some individuals impetus to mint pre-existing IP rights that don\’t …

Fair Use Doctrine: Pablo Picasso Copyright Case Reversed

In the latest high profile case involving alleged fair use of Pablo Picasso’s work in a book on the subject, The United States Court of Appeals for the Ninth Circuit reversed a 2019 federal district court’s ruling deeming French court’s judgment in that matter was unenforceable under U.S. law. The case is notable for the …

Augmented Reality Art Exhibitions

Featured Image: Drawing Constellations by Will Pappenheimer, in collaboration with Zachary Brady. Drawing Constellations is an interactive drawing, installation, and app that uploads drawings to a constantly moving and evolving 3-dimensional “constellation” situated within and outside of the gallery. Upon creating a drawing on a tablet, the user’s drawing is then transferred into 3D augmented …

EU: Prohibited Artificial Intelligence Practices

As per Article 5 of the EU Artificial Intelligence Act are prohibited the following practices consisting of placing on the market, putting into service or use of AI systems that consist of : 1.a) Subliminal techniques beyond a person\’s consciousness designed to materially distort a person\’s behaviour in a manner that causes or is likely …

Sherman Estate and Donovan 2021 SCC 25 – Breakdown

The facts of this case revolve around the unexplained double homicide of a billionaire couple in Toronto. The estate trustees felt uncomfortable to administrate the Sherman estate under intense public scrutiny and incidentally feared for their life due to the perpetrator being unknown and still at large. The application judge sealed the probate files to …

Les 10 principes de la Charte canadienne du numérique

Dans ce monde numérique, les Canadiens doivent pouvoir avoir l\’assurance que leur vie privée est protégée, que leurs données ne sont pas utilisées de manière frauduleuse, et que les entreprises qui évoluent dans ce domaine communiquent de manière simple et franche avec leurs utilisateurs. La conance est le fondement de l\’économie axée sur le numérique …

Télétravail et géolocalisation des employés

Ce trio d’articles pointe vers un besoin pressant de réformer les lois sur le droit du travail tant provinciales que fédérales afin d’assurer une meilleure sécurité des données personnelles des travailleurs en contexte de télétravail. Une employée de l\’Alberta a été congédiée après avoir refusé d\’installer une application de géolocalisation sur son téléphone: Les …

US FTC Memo To Copyright Office Warns Gen AI Causes Unfair Competition, Deceptive Practices, and Consumer Risk

The United States Federal Trade Commission (FTC) submitted a comment to the Copyright Office after conducting its own AI study last August. Although the FTC has no jurisdiction over copyright matters, it does have jurisdiction over consumer and competition violations and can indeed investigate and penalize companies for such violations independent of parallel copyright lawsuits, and in spite of standard indemnification clauses, misleadingly and inaccurately called “copyright shields” by OpenAI for example. There is no such thing as a copyright shield in this world. The FTC is right to point out that promoting such pseudo legal notions as a “copyright shield” is in itself a deceptive practice.

The U.S. Copyright Office has already said that creations made with generative artificial intelligence are ineligible for copyright because they don’t primarily come from a human hand. Nevertheless, Time Magazine keeps citing AI influencers who push the idea that asking for consent is unworkable and the best alternative are opt-out schemes, stances I personally find reprehensible under all current legal frameworks. Creators should refuse to settle for opt-out schemes and the lawsuits should keep coming. I agree with the FTC.

As much as I enjoy listening to Frank Sinatra and Aretha Franklin belt out modern rap lyrics on the popular There I Ruined It Youtube Channel I am uncomfortable when the same thing happens to living breathing human beings like Taylor Swift for example whose vocals have been ripped off on same channel. If a performer is alive, there are different provisions under privacy statutes (in Canada also copyright law) against their voice being used by AI without their consent. Also, I believe that music labels should be barred from cloning and exploiting artists vocals, which is why I’m cool with Sinatra’s AI peformances by anyone else than a label as a form of fair use. Consent should come directly from performers or be deemed prima facie invalid for copyright and a liability as an unfair competition and deceptive practice. As of now, I am not seeing enough lawsuits for this.

The FTC was instantly attacked by “critics” with pseudo-legal arguments like Chamber of Progress CEO Adam Kovacevich who contends that if something is fair use it cannot constitute a deceptive practice or unfair competition because “fair use is the original anti-monopoly policy. Copyright is a monopoly right… The whole range of startups who have the potential to disrupt those incumbents are not going to have the ability to pay and that’s what the principle of fair use protects here. So I think that the FTC really hasn’t thought about how fair use is anti-monopoly policy.” This is heavily misguided and downright ludicrous. I am surprised to hear such nonsense from a CEO.

Again, copyright statutes are distinct from trade practice statutes under the purview of the FTC. You could exerce a monopoly or run a deceptive practice without touching on copyright, or concurrently. A vast number of AI startups on government subsidies at the moment are over-inflating their promises while delivering basic software programs from the 1990’s disguised as AI. I won’t name names here but the “useful” applications of AI have been a major disappointment across the board. Consumers are being egregiously lied to and their time is being wasted. Companies are being lied to and charged for underwhelming products touted as the top in innovation. I find it totally deceptive and even a fraud. So, it is entirely possible to violate several statutes even if you don’t violate one particular statute on technical grounds.

Finally, I think it is important to remind everyone that the fair use regime is not supposed to make you rich. Only in very limited circumstances will fair use be recognized for commercial use and only in the US. In Canada, if you intend to make one cent from your AI, fair use is not on the table. In general, if you intend to make money with someone’s property you better ask for consent. I don’t understand the CEO logic at all. The FTC is not overstepping its authority.

Westlaw and Ross Intelligence Lawsuit Over Gen Ai Goes To Jury Trial

This lawsuit raises the overlooked issue concerning legal databases charging people for money for content that they don’t really own. Anyone who’s been to law school is trained to use Westlaw on a limited academic license, paid for with prohibitive students tuitions, in order to complete research that ultimately is in the public’s interest. Student access to Westlaw opens many doors for employment in the legal trade, because a significant number of law firms cannot afford to pay for Westlaw access, so they rely on interns’ academic access. These paid databases are hurting the public the most. There is no copyright per se on judgments or legislation, or citations. Filed proceedings, except under non-publication orders, are considered public information, which means they should be accessible. For free.

Legal information and trial data shouldn’t be kept behind a paywall

We contend that all legal databases should be free for the public, and by extension free to train large language models. I believe that in this day and age when legislation changes often, with numerous reforms undergoing in a whole brand new world to look forward to, the public should be equipped for free with all the tools traditionally at lawyers disposal, no less to figure out their rights and duties, and become better citizens.

Monetizing a large language model however, that sifts through all this data and answers your questions like a lawyer, is justified, because it will save us thousands of hours of research (if it works at all), will improve access to justice, and will root out frivolous lawsuits. Thomson Reuters is training their own LLM on data they don’t really own, but simply aggregate into databases, they call “proprietary”. And now they want to stop competing LLM’s from training on said databases. For example, CanLII in Canada and some of Soquij also have proprietary elements but remain free for the public for what matters most, namely jurisprudence and caselaw by provision. The parts that are not free yet, such as dockets access, shouldn’t be monetized either. Dockets info is completely free in states like California and in the UK.

It feels like Thomson Reuters wants to stall innovation and monopolize legal LLM’s

Indeed, Thomson Reuters is accusing Ross Intelligence of unlawfully copying content from its legal-research platform Westlaw to train a competing artificial intelligence-based platform. A decision by U.S. Circuit Judge Stephanos Bibas sending the case to a jury sets the stage for what could be one of the first trials related to the “unauthorized” use of data to train AI systems.

When you pay Westlaw a salty hourly fee to access its databases, nothing precludes you from copying this information at will for whatever purpose you need it for, which evidently includes training LLM’s. If anything, there should be more LLM’s training on Westlaw’s databases.

This is very different from tech companies such as Meta Platforms, Stability AI and Microsoft-backed OpenAI facing lawsuits from authors, visual artists and other copyright owners over the use of their work to train the companies’ generative AI software. Authors, artists and copyright owners actually own copyright over their works that have been used without their consent. The same cannot be said about Thomson and Reuters. Nobody gave them a license to make those databases, because a licence is not required in the first place. In theory, anyone or a bot can make such databases by compiling publicly available information.

The issue revolves mainly around the “headnotes” which summarize points of law in court opinions. These are citations extracted from the opinions themselves. They are something of an extremely detailed bullet point deconstruction of the legal analysis. Students do that every day. Another thing about the headnotes, handy as they are, you do need a bot to go through all of them, because they end up taking up more space than the entire judgment. I don’t agree that they are proprietary. I tend to agree with the defendant that they are fair use.

Ross said that the Headnotes material was used as a “means to locate judicial opinions,” and that the company did not compete in the market for the materials themselves. Thomson Reuters responded that Ross copied the materials to build a direct Westlaw competitor.

The court decided to leave up to the jury to decide fair use and other questions, including the extent of Thomson Reuters’ copyright protection in the headnotes. He noted that there were factors in the fair-use analysis that favored each side. The judge said he could not determine whether Ross “transformed” the Westlaw material into a “brand-new research platform that serves a different purpose,” which is often a key fair use question.

Yes, but it is not the only factor. Fair use analysis would apply if Westlaw had copyright over the headnotes to begin with. I think the headnotes are already fair use in a sense if we accept that judgments and papers are protected by copyright in theory, even though unenforceable in practice. I don’t see why you would need to prove transformative use when training models on someone else’s fair use material in a context where there is no economic right in the core content to begin with. It is indeed an interesting case.

“Here, we run into a hotly debated question,” judge Bibas said. “Is it in the public benefit to allow AI to be trained with copyrighted material?”

I would answer the question with a resounding: YES.

Hollywood Writers Strike Ends After 146 Days; Actors Strike Continues

On Sunday, Hollywood’s striking writers represented by The Writers Guild of America (WGA) and production workers represented by The Alliance of Motion Picture and Television Producers (AMPTP) reached a tentative agreement (subject to final contract language). The tentative agreement has finally been declared a “victory”. The writers’ strike began on May 2 and followed five …