This Week in Tech (Audio)

TWiT 1074: Chicken Mating Harnesses - Supreme Court Rules AI Art Not Copyrightable

March 9, 2026

Key Takeaways Copied to clipboard!

  • The Pentagon officially designated Anthropic as a supply chain risk following the company's refusal to allow its AI to be used for domestic surveillance or autonomous killing, setting a precedent for government control over AI development boundaries. 
  • The current frontier AI industry is economically incoherent, burning massive capital with no clear path to profitability, leading Cory Doctorow to predict a crash far worse than 2008. 
  • The Supreme Court's refusal to review the case confirms that AI-generated art lacks copyright protection because human authorship remains a bedrock requirement for copyright law. 
  • The U.S. Supreme Court's refusal to review the case confirms the current legal stance that AI-generated art, lacking human authorship, cannot be copyrighted, even if the prompt creator claims authorship. 
  • The debate over AI-generated software licensing is highlighted by the Chardet dispute, where a maintainer relicensed a derivative work under MIT after a clean-room rewrite using Claude code, challenging traditional viral open-source licensing like LGPL. 
  • The practice of using AI to analyze and rewrite copyrighted works, even through 'clean room' style processes involving LLMs trained on that code, is legally questionable and undermines established open-source licensing principles. 
  • The discussion touches on the broader implications of proprietary software control, noting that US trade policy historically enforced anti-circumvention laws to prevent modification and repair, a dynamic now being challenged globally due to geopolitical shifts and the capabilities of AI. 
  • High-growth tech companies rely on speculative future markets (like AI or Web3) to maintain high stock valuations, as mature companies with high market share cannot easily demonstrate growth necessary for favorable price-to-earnings ratios. 
  • Non-disparagement clauses, often included in employment contracts, can be used by powerful companies like Facebook to silence critics, even former employees or regulators, through severe financial penalties. 
  • The move toward privacy-focused mobile operating systems like GrapheneOS on non-Pixel hardware (like Motorola) is gaining traction as users seek alternatives to the increasingly locked-down Android and iOS ecosystems. 

Segments

AI in Warfare and DoD Conflict
Copied to clipboard!
(00:05:00)
  • Key Takeaway: The Pentagon officially designated Anthropic a supply chain risk after the company refused to allow its AI for surveillance or autonomous killing, prompting a presidential directive to cease all federal use.
  • Summary: Anthropic drew a ‘bright red line’ against using its AI for surveilling Americans or autonomous killing, which the Pentagon countered by declaring the company a supply chain risk. This action effectively bars other government agencies from using Anthropic’s technology, despite the company’s ongoing $200 million contract. Critics argue that if AI is a weapon, it should be regulated as one, while others point out that private companies should not dictate military AI use.
AI Ethics and Corporate Pretexts
Copied to clipboard!
(00:09:10)
  • Key Takeaway: The ethical stances of AI companies like Anthropic are viewed skeptically, as their stated ‘bright lines’ only prohibit certain uses ‘just not yet,’ suggesting a lack of fundamental ethical commitment.
  • Summary: Corey Doctorow notes that Anthropic’s refusal is conditional, implying they would use the technology if it were ‘better at it,’ which undermines the idea of having fixed ethical boundaries. The discussion highlights that the difference between companies like OpenAI and Anthropic might only be pretexts rather than fundamental ethical disagreement regarding military applications. The concept of ’transitive cooties’ describes how the supply chain designation makes any civilian organization using the service suspect.
AI Bubble Economics and Crash Prediction
Copied to clipboard!
(00:30:07)
  • Key Takeaway: The AI gold rush is economically unsustainable because current models lose more money with every use, accelerating losses rather than generating productivity gains like previous tech booms.
  • Summary: Unlike the dot-com era where every user made the web less unprofitable, every use of current AI accelerates the rate at which companies lose money, with sector revenue ($60B/year) vastly insufficient to cover CapEx ($600-700B) amortized over short hardware lifecycles. Investors are reportedly being misled by the sheer amount of money being spent, mistaking high CapEx for high upside, a dynamic Cory Doctorow predicts will lead to a crash making 2008 look minor.
Worker Autonomy vs. Automation
Copied to clipboard!
(00:54:13)
  • Key Takeaway: The value derived from AI tools depends entirely on worker autonomy; when humans are treated as ‘reverse centaurs’—peripherals used up by the machine—productivity is prioritized over quality, contrasting with skilled professionals using AI as true assistants (‘centaurs’).
  • Summary: A ‘centaur’ is a human assisted by a machine (like using a spell checker), while a ‘reverse centaur’ is a human used up by the machine to maximize throughput, exemplified by warehouse packers. For highly skilled workers like mathematicians, using AI tools like Claude allows them to maintain autonomy and achieve high-quality results by choosing when and how to integrate the tool. Conversely, corporations aim to use AI to reduce radiologists to ‘moral crumple zones’ who rubber-stamp outputs while bearing the liability.
AI Art Copyright Ruling
Copied to clipboard!
(01:02:41)
  • Key Takeaway: The Supreme Court declined to review the case, solidifying the ruling that AI-generated art lacks human authorship and is therefore not copyrightable.
  • Summary: Stephen Thaler attempted to copyright an AI-created image on behalf of the AI, but the Copyright Office and subsequent courts maintained that human authorship is a bedrock requirement. Declining review signals the Supreme Court’s unwillingness to hear similar cases, effectively confirming the current legal precedent. Copyright is automatic upon creation, but registration is necessary to access substantial statutory damages.
Prompting vs. Creation Analogy
Copied to clipboard!
(01:03:37)
  • Key Takeaway: Describing desired output via a prompt is analogous to commissioning a piece of art, not being the creator.
  • Summary: Prompting an AI is compared to a Renaissance duke commissioning a local artist, suggesting the prompter is merely commissioning work. The courts have suggested that creativity might reside in the prompt itself, which could potentially be copyrighted. Copyright protection is based solely on creativity, not on the labor or effort expended.
Geopolitical Tech Policy Shift
Copied to clipboard!
(01:08:19)
  • Key Takeaway: US-dominated tech policy for decades constrained global tech policy via trade agreements requiring weak privacy laws and banning reverse engineering.
  • Summary: For generations, free trade with the US required foreign partners to adopt weak privacy enforcement and anti-circumvention laws that banned modifying or reverse-engineering American products. This structure protected high-margin business models by locking consumers into consumables like printer ink. Recent US actions, including cutting off government access to Microsoft 365 for political adversaries, are causing global entities to seek data localization and jailbreak American platforms.
Chardet License Dispute
Copied to clipboard!
(01:15:35)
  • Key Takeaway: A maintainer relicensed the Chardet library under MIT after a clean-room rewrite using Claude code, challenging the viral nature of the original LGPL license.
  • Summary: The maintainer of the Python library Chardet created a new version using Claude code, claiming it was a qualitative rewrite allowing an MIT license instead of the required LGPL. The Free Software Foundation noted that using an LLM trained on the original code prevents it from being a true ‘clean room’ implementation, unlike historical examples like the Phoenix BIOS rewrite. Bruce Perens argues this development signals the death of software economics because all AI training is copying and all output is copying.
Grammarly Expert Review Criticism
Copied to clipboard!
(01:27:09)
  • Key Takeaway: Grammarly’s ’expert review’ feature, which suggests writing improvements based on trained authors without permission, is criticized as a gross misapprehension of writing and style.
  • Summary: Grammarly is using the works of writers like Cory Doctorow and Stephen King to inspire suggestions, claiming the output is only ‘inspired by’ their styles. This practice is likened to writer karaoke, where the output lacks the numinous, irreducible feeling inherent in human art or advice. The core malpractice is suggesting that an LLM trained on an author’s corpus can accurately advise on an individual’s unique writing voice.
Art, Soul, and AI Output
Copied to clipboard!
(01:33:07)
  • Key Takeaway: Art requires infusing a complex, numinous feeling from the artist’s mind into a vessel, a process AI cannot replicate, resulting in hollow, statistically averaged output.
  • Summary: True art involves transferring an irreducible feeling from the artist’s mind into a medium for another person to experience a facsimile of that feeling. AI models, by definition, know nothing of the user’s numinous feeling, making the output filler that converges on a bland statistical average, which is why AI art often feels hollow. This process is contrasted with human memory, which is lossy but still involves a unique, non-replicable internal process.
AI Growth Curve and Investment Hype
Copied to clipboard!
(01:43:32)
  • Key Takeaway: The rapid advancement of LLMs may represent a rich seam that will eventually plateau, similar to past computing techniques, but the current hype is fueled by the need for high P/E ratios to attract investment.
  • Summary: New computer science techniques often yield high dividends quickly before hitting a plateau where new techniques are required, suggesting LLMs might follow this pattern. Companies with high market share need growth stories to maintain high price-to-earnings ratios necessary for stock-based acquisitions. The narrative of creating ‘God’ or ‘mind control’ serves as ‘criti-hype’ to attract massive investment capital, regardless of the actual technological limits.
Vein Finders and Medical Tech
Copied to clipboard!
(01:47:03)
  • Key Takeaway: Vein finder devices, which use light absorption to project real-time vein maps onto the skin, are technologically superior to older methods like X-ray shoe fittings.
  • Summary: Vein finders operate by shining a specific light spectrum absorbed by blood, projecting black lines where veins are located, offering a Star Trek-like utility. This technology is vastly safer than historical practices like using fluoroscopes in shoe stores, which irradiated customers’ faces. The device, exemplified by the ‘Hello Vane’ model, provides real-time visualization of vasculature.
Meta Glasses Privacy Breach
Copied to clipboard!
(02:02:23)
  • Key Takeaway: Meta’s Ray-Ban smart glasses are sending recorded images to subcontractors in Kenya for model training, leading to contractors viewing intimate user moments.
  • Summary: Meta subcontractors in Nairobi are labeling and annotating data captured by the smart glasses, including intimate or private moments, because users forget the devices are recording. This issue is a recurring pattern across smart speakers and voice assistants where ambiguous or failed commands are offloaded for human review. The lack of object permanence regarding privacy risks is a persistent failure in consumer technology adoption.
Growth Stock Valuation and AI Hype
Copied to clipboard!
(02:05:37)
  • Key Takeaway: Speculative future markets like AI are necessary for tech companies to maintain high price-to-earnings ratios and use stock for acquisitions.
  • Summary: Companies with high growth potential, like those in the AI rush, receive market valuations of 20x to 200x earnings, allowing them to issue stock for acquisitions. Mature companies with near-market saturation, like Google’s search dominance, must invent non-existent markets (like Web3 or AI superintelligence) to avoid being revalued as mature stocks with lower market caps. This reliance on the ‘imaginary’ prevents market correction until the promised future materializes or fails to materialize.
Google Epic Games Settlement Details
Copied to clipboard!
(02:08:29)
  • Key Takeaway: The Google-Epic Games settlement includes a six-year non-disparagement clause for Epic’s Tim Sweeney.
  • Summary: Google and Epic Games settled their rivalry, resulting in Google ending its 30% app store fee for certain metaverse apps and welcoming third-party stores under pressure from the EU. A key term requires Tim Sweeney not to disparage Google for six years. This is contrasted with author Sarah Wynn Williams, who faces an $111 million fine from Facebook for violating a similar clause, highlighting the punitive nature of these agreements.
Non-Disparagement Clause Enforcement
Copied to clipboard!
(02:10:23)
  • Key Takeaway: Non-disparagement clauses can be legally binding even for high-level regulators, creating conflicts of interest.
  • Summary: Employment contracts often contain non-disparagement agreements, which can be enforced even against whistleblowers, depending on state law exceptions. The top privacy regulator in Ireland, an ex-Facebook executive, is understood to be bound by such a clause, preventing criticism of Facebook while regulating them. State legislatures have the power to deem certain clauses, like non-competes in California, unenforceable as a matter of public policy.
Xbox Next-Gen Console Details
Copied to clipboard!
(02:12:55)
  • Key Takeaway: The next-generation Xbox, codenamed ‘Project Helix,’ will support PC games alongside console titles.
  • Summary: The Xbox CEO confirmed the next console, ‘Project Helix,’ will be capable of playing PC games. This development echoes the history of the original Xbox jailbreak by Bunny Wang, who sought to enable PC game compatibility. Bunny Wang is now working on an entirely open-source mobile platform called Precursor to combat hardware supply chain attacks.
GrapheneOS Motorola Partnership
Copied to clipboard!
(02:16:28)
  • Key Takeaway: Motorola is partnering to support GrapheneOS, offering the first non-Pixel phone running a Google-less Android version.
  • Summary: Motorola’s support for GrapheneOS means consumers can purchase a stock Android phone running the open-source, Google-free version of Android. This is significant because Google is expected to lock down the Pixel bootloader, making GrapheneOS installation difficult on future Pixels. Installing GrapheneOS currently requires wiping the existing phone data, making the transfer process a major hurdle for non-technical users.
Twitter API Block on Mastodon Growth
Copied to clipboard!
(02:19:48)
  • Key Takeaway: Elon Musk killed the virtuous cycle of Mastodon growth by blocking third-party apps that identified and auto-followed Twitter contacts moving to Mastodon.
  • Summary: The ability for users to easily migrate from Twitter to Mastodon was severely hampered when Twitter blocked apps that cross-referenced contacts between the platforms. This action disrupted the organic growth cycle where users would naturally follow their existing network as they migrated. The discussion also touched on Blue Sky’s binding arbitration clause, which discourages adoption by privacy-conscious users.
Data Broker Breaches and Privacy Laws
Copied to clipboard!
(02:28:55)
  • Key Takeaway: Congress prioritizes protecting its own privacy via laws like the Doxing ban while neglecting comprehensive consumer privacy legislation.
  • Summary: Data broker breaches have resulted in nearly $21 billion in identity theft losses, yet federal consumer privacy laws have stalled since 1988. Congress recently passed a law banning doxing of federal lawmakers, which Senator Ron Wyden opposed because it did not protect the general public or state-level officials. Law enforcement agencies favor data brokers as a cheap alternative to obtaining data via warrants.
Government Use of Ad Tracking Data
Copied to clipboard!
(02:32:54)
  • Key Takeaway: Customs and Border Patrol (CBP) utilizes location data purchased from the online advertising ecosystem to track individuals’ movements.
  • Summary: A DHS document revealed that CBP uses location data sourced from the online advertising industry to track phone locations, confirming fears about the misuse of amassed personal data. This practice validates earlier fictional accounts predicting the use of ad tech data for government surveillance. The discussion also noted that running ‘chaff’ scripts (generating random search activity) results in bizarre targeted ads, such as for chicken mating harnesses.
ProtonMail Identity Disclosure
Copied to clipboard!
(02:36:04)
  • Key Takeaway: ProtonMail assisted the FBI in unmasking a ‘Stop Cop City’ protester, demonstrating it is a privacy tool, not an anonymity service.
  • Summary: ProtonMail cooperated with the FBI to reveal a user’s identity, arguing they protect communication integrity but must comply with warrants regarding user identity. This outcome is foreseeable when companies establish personnel or assets in jurisdictions where they can be coerced, despite being based in Switzerland. The segment concluded that email, even encrypted, should not be considered a private function due to visible metadata and the difficulty of widespread PGP adoption.
COPPA 2.0 and Age Verification
Copied to clipboard!
(02:40:11)
  • Key Takeaway: The updated COPPA 2.0 unanimously passed the Senate despite including a provision that exempts data collection for age verification.
  • Summary: The Children and Teens Online Privacy Protection Act (COPPA 2.0) passed the Senate unanimously, but it contains a loophole allowing companies to collect ID information for age verification without being subject to the act’s restrictions. This mirrors the pattern of collecting massive amounts of sensitive data, which inevitably leads to future security risks. The evidence base used to justify the legislation is considered weak and grossly overstated.
South Korean Crypto Wallet Loss
Copied to clipboard!
(02:41:35)
  • Key Takeaway: South Korean authorities lost $5.6 million in seized cryptocurrency after publishing high-resolution photos of hardware wallets displaying the seed phrases.
  • Summary: South Korean tax authorities compromised a seized hardware wallet containing millions by publicly posting images of the device that clearly showed the wallet’s seed phrase. This incident underscores the critical importance of protecting physical recovery information, as high-quality images can facilitate immediate theft. The segment noted that the better the camera technology, the better the hack.
Iranian Drone Strike on Amazon Data Centers
Copied to clipboard!
(02:42:49)
  • Key Takeaway: Iranian drones struck Amazon data centers in the UAE and Bahrain, contradicting the expectation that basing operations in Dubai insulates businesses from geopolitics.
  • Summary: Amazon data centers in the UAE and Bahrain were physically hit by objects falling from the sky, identified as Iranian drones. This event undermines the perceived geopolitical insulation offered by establishing operations in locations like Dubai. The incident raises alarms about the physical security of cloud infrastructure in volatile regions.
ISP Consolidation and Charter/Cox Merger
Copied to clipboard!
(02:51:37)
  • Key Takeaway: The proposed merger of Charter and Cox, which would create the largest US ISP, faces state-level opposition despite FCC approval.
  • Summary: Charter is seeking FCC permission to acquire Cox, which would give them 35.6 million customers, surpassing Comcast. The deal requires sign-off from the Justice Department and state regulators in California and New York, where opposition is expected. Charter’s former CEO was criticized for poor pandemic-era treatment of technicians, including giving them vouchers for restaurants that were closed due to lockdowns.
Big Diaper Market Squeezing Tactics
Copied to clipboard!
(02:53:29)
  • Key Takeaway: Diaper manufacturers allegedly push back toilet training ages to grow the market after achieving near-monopoly status.
  • Summary: After dominating the market, diaper companies allegedly seek marginal gains by extending the period children use diapers, pushing the average toilet training age from 18 months in 1947 to 37 months by 2004. This tactic is analogous to Google allegedly lowering search quality to increase ad impressions when market growth stalls. The industry is now expanding by offering larger sizes, including size eight diapers for children up to 65 pounds.
23andMe Revival and Junk Science
Copied to clipboard!
(02:56:54)
  • Key Takeaway: 23andMe is attempting a revival focused on ‘rich donors’ and ‘improved tests,’ despite its history of pseudoscience regarding genetic ancestry.
  • Summary: Anne Wojcicki’s plan to revive 23andMe involves securing wealthy donors and offering enhanced tests, potentially including personalized medicine claims. The company’s ancestry reporting, which assigns percentages like ‘17% Viking,’ is criticized as pseudoscience based on comparing genomes against a self-selected group of ‘real Germans.’ The risk is compounded because the parents of current users already provided their genetic data, making the data non-consensually available.
Human Neurons Playing Doom
Copied to clipboard!
(02:59:05)
  • Key Takeaway: Cortical Labs demonstrated 200,000 living human neurons playing the game Doom at a beginner level.
  • Summary: A biological computer, the CL1, utilizing 200,000 human neurons grown on an electrode array, was shown to interact with the game Doom. The performance level was equivalent to a complete beginner who had never seen a computer before, suggesting the signals were largely random. This development highlights progress in bio-computing, even if the current application is rudimentary.
Firefox Crashes from Bit Flips
Copied to clipboard!
(03:00:20)
  • Key Takeaway: Approximately 10% of Firefox crashes are directly attributable to memory bit flips, indicating hardware instability.
  • Summary: Data collected from Firefox crash reports and memory testing confirms that a significant portion of crashes stem from users having faulty or non-ECC RAM. This validates Linus Torvalds’ long-standing recommendation for using error-correcting memory in critical systems. The discovery suggests that cheap, low-quality hardware components can introduce substantial software instability.
Seagate’s 44TB Hammer Drives
Copied to clipboard!
(03:01:54)
  • Key Takeaway: Seagate has released 44TB hard drives utilizing Heat-Assisted Magnetic Recording (HAMR) technology, with 100TB drives anticipated.
  • Summary: Seagate unveiled 44-terabyte, three-and-a-half-inch hard drives using HAMR technology, primarily targeting hyperscalers and large AI data storage facilities. The technology involves using heat to aid magnetic recording, which raises concerns about data integrity given the Firefox bit-flip statistics. These high-capacity drives are driving innovation in large-scale AI data storage solutions.