← the graph
Sector 6 · 8 sub-themes

Safety, Alignment & Policy

The frontier is now a national-security battleground. Anthropic told the Senate Banking Committee that Alibaba's Qwen division ran the largest "adversarial distillation" attack in its history, harvesting Claude's software-engineering and agentic reasoning to train a rival. Washington treated frontier models as strategic assets in kind. It forced Anthropic to shut Claude Fable 5 roughly three days after its June 9 release, then permitted Mythos 5 only to about 100 vetted companies and agencies; the controls were lifted after a 19-day ban. OpenAI got a parallel order to throttle GPT-5.6, and Altman conceded the company "made commitments." Legal tech firm Legion is already suing over the restrictions. Copyright litigation is escalating on a second front. Roughly 400 local newspapers sued OpenAI and Microsoft, the NYT amended its suit, and Suno, Apple, and Anthropic all face training-data claims. Governance is racing to catch up. At the UN's Geneva summit, Bengio and Ressa warned oversight is outpaced, while Illinois became the first US state to mandate third-party frontier AI safety audits. INTERPOL flagged a surge in AI-driven phishing and deepfake executive-impersonation scams across Asia-Pacific. Meanwhile creative-industry backlash is sharpening: Netflix cloned Gene Wilder's voice, Hasbro demanded likeness rights from Peppa Pig child actors, and A24's Google DeepMind deal drew fan revolt. The UK, notably, chose investment over restriction, funding the £40M BOLD research lab.

AnthropicOpenAIAlibabaMicrosoftLegionTim ScottElizabeth WarrenYoshua BengioMaria RessaSam AltmanQwenClaude Fable 5national security AI export controlsadversarial distillation attacksAI copyright litigationfrontier model regulationUS-China AI competitionAI governance fragmentationdeepfake cybercrimeAI voice and likeness rightsstate-level AI auditsstrategic AI asset control
6.1

Anthropic Accuses Alibaba of Large-Scale Distillation Attack on Claude

  • In a letter dated June 10, 2026, addressed to U.S. Senate Banking Committee Chair Tim Scott and Ranking Member Elizabeth Warren, Anthropic alleged that operators affiliated with Alibaba and its AI research division, Alibaba Qwen, ran a coordinated "adversarial distillation" campaign from April 22 to June 5, 2026, generating more than 28.8 million exchanges with Claude through nearly 25,000 fraudulent accounts. [8][1][5]
  • The campaign specifically targeted Claude's most advanced capabilities — software engineering and agentic reasoning — described by Anthropic as the core of its Mythos Preview model; Anthropic described the technique as one where a less capable AI is trained on a more powerful model's outputs to replicate frontier capabilities without incurring the R&D and computational costs of training from scratch. [8][6]
  • Anthropic characterized this as the largest known distillation attack in its history, warning U.S. senators and White House officials that Chinese AI labs are harvesting U.S. frontier AI capabilities at industrial scale to repackage them as their own. [8][2][7]
  • Separately, Anthropic faces a $75 million lawsuit from approximately 100 authors who allege the company pirated their books to train its Claude AI models. [3][4]
AnthropicTim ScottElizabeth WarrenAlibabaAlibaba QwenClaudeMythos PreviewU.S. Senate Banking CommitteeWhite Houseadversarial distillation attackAI capability theftfrontier model piracyChinese AI labsagentic reasoningsoftware engineering AItraining data scrapingcopyright lawsuitU.S.-China AI competitionindustrial-scale model distillation
Sources
  1. [1]Ana Maria Constantin — The AI company told US senators and White House officials that operators linked to Alibaba's Qwen lab used nearly 25,000 fraudulent accounts★news
  2. [2]abharade@insider.com (Aditi Bharade) — Anthropic accused one of China's biggest tech companies of exploiting its advanced AI models. Anthropic's head of policy, Sarah Heck, said i★news
  3. [3]BeInCrypto — Anthropic Faces a New $75 Million Lawsuit for Pirating Books to Train Claude AI★news
  4. [4]New York Post — 100 authors demand $75M from Anthropic over ‘stolen’ work to train its systems: lawsuit★news
  5. [5]kimmonismus — Anthropic claims: Alibaba continues to distill Claude on a large scale to train Qwen. Via Bloomberg Anthropic is accusing Alibaba-linked ope★twitter
  6. [6]DataChaz — 🚨 Anthropic is accusing Alibaba, DeepSeek, and other Chinese tech giants of orchestrating the largest AI heist in history. Over 45 days, the★twitter
  7. [7]Reuters — Anthropic has accused Alibaba, the Chinese technology and e-commerce giant, of illicitly extracting its Claude AI model capabilities in what★twitter
  8. [8]The_Cyber_News — 🧑‍⚖️Anthropic Accuses Alibaba of 'Illicitly' Accessing Its Claude AI Models Source: https://t.co/qw1p2rWzZT Anthropic has formally accused C★twitter
6.2

U.S. export controls hit Anthropic and OpenAI frontier AI models

  • The U.S. government, citing national security authorities, forced Anthropic to shut down access to Claude Fable 5 approximately 3 days after the model's release on June 9, making it one of the shortest public availability windows for a major AI model. [1][6][10]
  • After initial talks between Trump administration officials and Anthropic failed to lift the export controls, the U.S. government subsequently granted Anthropic permission to release its Claude Mythos 5 model to a restricted group of roughly 100 companies and federal agencies. [3][4][8]
  • The U.S. government also told OpenAI to limit the rollout of GPT-5.6 while the government reviews it; OpenAI CEO Sam Altman stated the company had "made commitments" in response. [9]
  • Legal tech company Legion filed the first legal challenge to the Trump administration's AI export controls targeting Anthropic's Fable 5, suing the administration over the restrictions. [7]
  • CNBC noted that Anthropic's longstanding advocacy for AI regulation has created complications for the company, now valued at close to $1 trillion, as government oversight it helped encourage is now being applied to its own model releases. [2]
  • One industry observer noted that the export control dynamic creates a strategic incentive for OpenAI to deliberately underperform on model releases to avoid triggering similar government restrictions and gain market share. [5]
AnthropicClaude Fable 5Claude Mythos 5OpenAIGPT-5.6Sam AltmanLegionTrump administrationCNBCAI export controlsnational security restrictionsmodel release shutdownsgovernment AI oversightlegal challenge to AI policyAI regulation advocacyrestricted model rolloutstrategic underperformance incentivefederal agency AI access
Sources
  1. [1]Cyber Magazine — Anthropicclaimed Claude Fable 5 was the most powerful cybersecurity model in the world when the firm released it on 9 June. However, the US★news
  2. [2]CNBC — Anthropic'slongstanding push for regulation is coming back to bite it. Theartificial intelligencecompany, now valued at close to $1 trillion★news
  3. [3]WIRED — Trump administration officialsconcluded talks with Anthropic on Monday without lifting export controls that wereimposed last weekon the comp★news
  4. [4]CNBC — The U.S. government on Friday grantedAnthropicpermission to release itsMythos 5model to a group of roughly 100 companies and federal agencie★news
  5. [5]dylan522p — OpenAI now just needs to sandbag the next model release So the US doesn't export control them So they gain massive market share It's imperat★twitter
  6. [6]minchoi — Claude Fable 5 lasted only 3 days before the U.S. government forced Anthropic to shut it down. People still built insane things up to its fi★twitter
  7. [7]kimmonismus — The first legal challenge to Trump’s AI export controls against Anthropics Fable 5is here. Legal tech company Legion is suing the Trump admi★twitter
  8. [8]wallstengine — The U.S. government LIFTED export blocks on Anthropic’s Claude Mythos 5 model, allowing access for 100+ trusted U.S. institutions, per Semaf★twitter
  9. [9]WatcherGuru — JUST IN: 🇺🇸 US tells OpenAI to limit GPT 5.6 release while the government reviews it, The Information reports. CEO Sam Altman: "we've made c★twitter
  10. [10]arena — The official statement from @AnthropicAI https://t.co/OtYB8vB0rM — quoting @AnthropicAI: The US government, citing national security authori★twitter
6.3

~400 Newspapers Sue OpenAI and Microsoft Over AI Training Copyright

  • The owners of approximately 400 local newspapers have filed a lawsuit against OpenAI and Microsoft, alleging that the companies used copyrighted journalism to train large language models without authorization or compensation. [3][4][5]
  • The New York Times has amended its existing lawsuit against OpenAI and Microsoft, reflecting ongoing and escalating legal disputes over the use of copyrighted news content in AI training. [1]
  • A Microsoft shareholder has separately sued the company's top executives over AI-related copyright claims, adding a corporate governance dimension to the broader wave of AI copyright litigation. [2]
  • Ottawa's recent AI strategy has drawn attention for not addressing AI systems' unauthorized use of copyrighted content, even as questions and lawsuits over that issue have mounted. [6]
OpenAIMicrosoftNew York TimesOttawaAI copyright litigationlarge language model training dataunauthorized use of journalismnewspaper publishers lawsuitcorporate governance AI liabilityAI policy gapscopyrighted content compensation
Sources
  1. [1]The New York Times — The New York Times Amends Lawsuit Against OpenAI and Microsoft★news
  2. [2]Bloomberg Law News — Microsoft Shareholder Sues Top Brass for AI Copyright Claims (1)★news
  3. [3]MorePerfectUS — OpenAI and Microsoft are being sued by the owners of 400 local newspapers nationwide for mass copyright infringement. The publishers claim t★twitter
  4. [4]WindowsCentral — Microsoft and OpenAI are facing a new lawsuit from publishers accusing them of using copyrighted news content to train AI models without pertwitter
  5. [5]MCO_News — Nearly 400 Newspapers Just Sued OpenAI and Microsoft The fight over who owns the data behind AI just got much bigger. Nearly 400 newspapers twitter
  6. [6]AnjaKaradeglija — Questions — and lawsuits — have swirled for years around AI systems’ unauthorized use of copyrighted content, but Ottawa's recent AI strategtwitter
6.4

AI governance debates span UN summits, state laws, and agentic risks

  • At a UN AI governance summit in Geneva, experts including Yoshua Bengio and Maria Ressa warned that faster AI progress is outpacing public oversight and independent testing; all 193 UN member states entered the debate, though a key concern raised was that most countries lack the compute needed to independently audit frontier AI systems. [3]
  • UN experts at the Geneva summit focused accountability discussions on concrete harms — deepfakes, online violence, and children's safety — raising the question of who is liable when AI harms people, connecting governance to human rights rather than abstract risk. [3]
  • Pre-summit analysis highlighted that any global AI governance agreement will be difficult to enforce without active cooperation from the countries and companies that control leading AI models and compute infrastructure, effectively making US-China alignment a prerequisite for meaningful global oversight. [3]
  • Illinois became the first US state to require outside AI safety audits, signing a frontier AI safety law that mandates large AI developers publish safety frameworks, report major incidents, and undergo annual third-party audits; because large firms typically comply with the strictest applicable state rule rather than build separate systems, the law could have national reach. [3]
  • Commentary from multiple sources notes that AI governance is increasingly framed as an infrastructure problem rather than an ethics or policy problem, with the EU AI Act and California's SB 53 cited as examples of regulation that rewrites operational infrastructure rather than just policy guidelines. [2]
  • The challenge of governing agentic AI systems is highlighted as a distinct and underaddressed problem: standard governance tools such as guidelines, access controls, and output reviews are described as necessary but insufficient when AI agents act autonomously across systems. [1][4]
Yoshua BengioMaria RessaUnited NationsIllinoisEU AI ActSB 53AI governanceglobal oversight frameworksfrontier AI safety auditsagentic AI regulationUN summitcompute infrastructuredeepfakes and online harmsUS-China alignmenthuman rights accountabilitystate-level AI legislation
Sources
  1. [1]youdotcom — Most teams treat AI governance as a policy problem. Guidelines, access controls, output reviews. Necessary, not enough. Real governance livetwitter
  2. [2]akavenetwork — AI regulation in 2025 stopped being about ethics. It became about infrastructure. The EU AI Act and SB 53 don't just rewrite policy. They retwitter
  3. [3]bigdata — https://t.co/BjfY2iUm6y 📡 Tracking the Costs, Risks, and Rules of AI https://t.co/b9t8hSnIS7twitter
  4. [4]cybersecboardrm — You Don't Have an AI Governance Problem. You Have an Agentic AI Problem https://t.co/ag2sYsU9Uk #AgenticAI #AIGovernance #Cybersecurity #Cybtwitter
6.5

INTERPOL warns of surging AI-driven cybercrime across Asia-Pacific

  • INTERPOL's 2025/2026 Asia and South Pacific Cyberthreat Assessment Report found a "dramatic increase" in cybercrime across the region, with phishing identified as the most widespread and financially damaging form — a third of countries in the region reported more than 10,000 phishing cases between January 2024 and March 2025, and over half of INTERPOL member countries reported that cybercrime accounted for at least 30% of all crimes recorded nationally. [1]
  • The same INTERPOL report documented more than 135,000 ransomware-related attacks in the region in 2024, alongside a surge in deepfake and AI-driven scams that impersonate business executives to authorize fraudulent transactions, with INTERPOL's Cybercrime Director Neal Jetton noting that cybercriminals are "leveraging artificial intelligence, ransomware-as-a-service models and sophisticated social engineering techniques on an industrial scale." [1]
  • Multiple cybersecurity commentators note that AI has shifted phishing attacks toward pure social engineering with no malware required, with one source citing that 90%+ of blocked attacks were social-engineering-only and referencing $1.28B in associated losses, while industry voices emphasize that countering AI-assisted threats requires combining AI-driven detection with human analyst oversight. [2][3]
INTERPOLNeal JettonAsia and South Pacific Cyberthreat Assessment Reportcybercrime surge Asia-Pacificphishing attacksransomware-as-a-serviceAI-driven scamsdeepfake fraudsocial engineeringbusiness email compromiseAI-assisted threat detectioncybercrime financial lossesINTERPOL cybersecurity report
Sources
  1. [1]Dinosn — INTERPOL Warns Phishing, Ransomware, and AI Scams Are Rising Across Asia-Pacific https://t.co/pdKnXjd6jjtwitter
  2. [2]GoLeadingIT — AI phishing doesn't look like the old obvious scams anymore. 90%+ of blocked attacks were pure social engineering, no malware needed. $1.28Btwitter
  3. [3]FITsolutionsIT — AI has transformed cybersecurity. But AI alone doesn’t stop breaches. Human-led cybersecurity combines: • AI-driven detection • Real analysttwitter
6.6

UK launches BOLD national AI research lab with £40M funding

  • The UK government launched a £40M call for a fundamental AI research lab, and due to "exceptional bids," funded two major new AI labs rather than one, with BOLD (British Open-Ended Learning and Discovery Lab) being one of them — a national academic lab focused on paradigm-shifting, open-ended learning and discovery research. [2][1]
  • BOLD consolidates five research labs from three world-leading UK universities under six co-directors, with its stated mission being to build a national capability in fundamental AI research in an academic setting. [4][6]
  • UCL DARK, built over eight years, announced it will sunset and merge into BOLD, described as an "upgrade" to its mission. [5][3]
UK governmentBOLDUCL DARKUCLUK AI research fundingfundamental AI researchopen-ended learningacademic AI labnational AI capabilityresearch lab consolidationparadigm-shifting AI research
Sources
  1. [1]bold_lab_ai — Hello world :) We are BOLD — the British Open-ended Learning and Discovery Lab! BOLD is a new academic research lab fully focussed on paraditwitter
  2. [2]KanishkaNarayan — 🚨 2 major new AI labs in the UK 🚨 3 months ago, we launched a £40m call for a fundamental AI lab in the UK. Given the exceptional bids we retwitter
  3. [3]_rockt — It has been an absolute privilege and pleasure to build up @UCL_DARK with @egrefen, @robertarail and @jparkerholder over the past eight yeartwitter
  4. [4]CULLYAntoine — Incredibly proud to be a co-director of @bold_lab_ai 🚀 Five outstanding labs, from three world-leading institutions, one shared mission — witwitter
  5. [5]UCL_DARK — We are excited to announce that we will sunset @UCL_DARK and merge into the British Open-Ended Learning & Discovery (BOLD) Lab. It's a utwitter
  6. [6]tonizza82 — Very excited and honoured to launch BOLD! 3 universities, 6 co-leaders, outstanding researchers, all together with one goal: building a natitwitter
6.7

AI voice cloning, child actor clauses, and AI ads spark backlash

  • Netflix's new Wonka series used an AI-generated recreation of the voice of deceased actor Gene Wilder, drawing significant public backlash over the ethics of digitally replicating a dead celebrity's voice without his direct consent. [4][5][6]
  • Hasbro introduced an AI clause requiring child actors on Peppa Pig to sign over rights to AI use of their voices and likenesses, prompting industry-wide backlash from parents, agents, and advocacy groups who characterized the practice as exploitative of minors. [7][8][9][10]
  • REI faced public backlash over suspicions that its advertising campaign used AI-generated imagery, with critics arguing the move contradicted the outdoor retailer's environmental and community-oriented brand values. [11]
  • SEGA clarified its use of AI in the development of Crazy Taxi: World Tour following controversy that emerged after the game's public reveal, though the company's specific clarifications addressed audience concerns about the nature and scope of that AI use. [2]
  • Meta introduced a paid monthly subscription tier for its AI features on smart glasses, drawing backlash centered on both the added cost and privacy concerns associated with AI-enabled wearable devices. [1]
  • The George Washington biopic Young Washington sparked generative AI backlash related to its use of AI-assisted production methods, drawing criticism from audiences and industry observers. [3]
NetflixGene WilderHasbroPeppa PigREISEGACrazy Taxi: World TourMetaYoung WashingtonGeorge WashingtonAI voice cloning of deceased celebritieschild actor likeness rightsAI-generated advertising imagerygenerative AI in entertainment productionAI ethics and public backlashsmart glasses AI subscriptiondigital replica consentexploitative AI contractsbrand authenticity concernswearable AI privacy
Sources
  1. [1]Tekedia — Meta Faces Backlash as New AI Subscription Adds Monthly Fee to Smart Glasses Amid Privacy Concerns
  2. [2]player.one — SEGA Clarifies 'Crazy Taxi: World Tour's' AI Use Following Controversy After Its Grand Reveal
  3. [3]Comic Basics — 'Young Washington' AI Controversy Explained, as the George Washington Biopic Sparks Generative Backlash
  4. [4]The Express Tribune — Netflix faces backlash over AI recreated Gene Wilder voice in new ‘Wonka’ series
  5. [5]BBC — Wonka Netflix show faces backlash for AI-generated Gene Wilder voice
  6. [6]MSN — Netflix's AI Gene Wilder voice for Wonka show draws backlash
  7. [7]Pajiba — ‘Peppa Pig’ Faces Industry Backlash for Controversial AI Contracts
  8. [8]Euronews.com — Peppa Pig controversy as Hasbro introduce AI clause for child actors
  9. [9]Deadline — 'Peppa Pig' Backlash As Hasbro Asks Child Actors To Sign AI Clause
  10. [10]PennLive.com — ‘Peppa Pig’ AI controversy receives backlash: ‘Pure [expletive] evil’
  11. [11]Fast Company — 'So much for caring about the environment': REI faces backlash over AI-generated ad suspicions
6.8

AI Copyright Disputes Intensify Across Music, Film, and Tech

  • Apple is actively defending its scraping of YouTube videos for AI training in a copyright lawsuit, arguing the practice is legally permissible. The case represents a notable legal test of whether large-scale video scraping for AI model training constitutes copyright infringement. [1]
  • Production music firm Jamendo has filed a new copyright lawsuit against AI music generator Suno, alleging unauthorized use of its catalog for AI training purposes. This follows earlier legal pressure on Suno from other music industry plaintiffs. [4]
  • Indie film studio A24 has entered into a partnership with Google DeepMind, triggering significant fan backlash from audiences concerned about the studio's embrace of AI. A24 publicly defended the partnership in response to the controversy. [5][6][7]
  • No Film School reports that the A24–Google DeepMind deal raises questions about what the collaboration entails for the studio known for prestige independent filmmaking, reflecting broader industry tensions around AI adoption in creative fields. [3]
  • A JD Supra client alert highlights the broader contradiction facing Hollywood, where studios simultaneously pursue copyright claims against AI companies for unauthorized training-data use while themselves adopting or partnering with AI tools — a tension the A24–Google DeepMind deal exemplifies. [2]
  • Google DeepMind has published material addressing the trajectory from AGI (artificial general intelligence) to ASI (artificial superintelligence), signaling the company's public framing of its long-term research direction. [8]
AppleYouTubeJamendoSunoA24Google DeepMindNo Film SchoolJD SupraAI training data copyright litigationmusic copyright infringementvideo scraping legalityAI partnerships in creative industriesHollywood AI adoption tensionsAGI to ASI research directionindie film and AI controversyunauthorized training data use
Sources
  1. [1]HotHardware — Apple Defends Scraping YouTube Videos For AI Training In Copyright Case
  2. [2]JD Supra — Client Alert: AI Copyright Showdown: Can Hollywood Challenge AI While Using It?
  3. [3]No Film School — What You Need to Know About A24's Partnership with Google DeepMind
  4. [4]Billboard — Suno Faces Another AI Training Copyright Lawsuit From Production Music Firm Jamendo
  5. [5]Futurism — A24 Is Dealing With a Full-on Crisis Among Fans After Signing an AI Deal With Google
  6. [6]Deadline — A24 Defends Google AI Partnership
  7. [7]WIRED — A24 Knows You’re Mad About the Google AI Collab
  8. [8]Google DeepMind — From AGI to ASI From AGI to ASI Google DeepMind