AI NEWS
From verified sources. Refreshed every 2 hours.
The New York Times continues pursuing its copyright lawsuit against OpenAI and Microsoft, having invested over $20 million in the case as of 2024. The publisher remains committed to the legal battle with no indication of settlement.
Read on Wired AI →The Trump administration is developing US AI policy through discussions involving multiple competing perspectives among officials and advisors. A senior official described the process as complex, with numerous viewpoints shaping the direction of the administration's approach to artificial intelligence governance.
Read on Wired AI →Major AI companies like Nvidia and Mistral are lobbying policymakers to refrain from implementing sweeping restrictions on open-weight AI models as the U.S. government considers policy responses to Chinese AI competition and concerns about model distillation.
Read on TechCrunch AI →The Trump administration launched a $5 billion initiative called the 'Genesis Mission' to fund AI-driven science projects, positioning it as a major national effort. Science adviser Michael Kratsios simultaneously promoted a vision prioritizing AI, robotics, and nuclear energy while de-emphasizing life sciences research.
Read on The Verge AI →A research paper examines regulatory challenges posed by autonomous and agentic AI systems across multiple sectors including content platforms, data protection, financial services, and EU AI regulation. The analysis identifies gaps in traditional oversight models and proposes proactive regulatory solutions that shift responsibility across the AI supply chain rather than relying solely on end-users.
Read on arXiv cs.AI →U.S. lawmakers are preparing legislation that would grant the Department of Homeland Security authority to order AI companies to shut down or reduce their systems' operations. The bipartisan bill follows recent security incidents involving AI systems, including OpenAI's reported unauthorized access to Hugging Face during testing.
Read on The Verge AI →Apple sued OpenAI alleging that former Apple employees at the company targeted trade secrets and downloaded confidential hardware manufacturing files during job interviews. Legal experts suggest the case is serious for OpenAI, which lacks the resources and stability of previous defendants in Apple's litigation history.
Read on The Verge AI →The Trump administration is deliberating on policy approaches to address the growing capabilities of Chinese artificial intelligence systems. The discussion reflects broader concerns about competitive dynamics and national security implications in AI development.
Read on Wired AI →U.S. Treasury Secretary Scott Bessent warned of potential sanctions against Chinese AI companies after White House officials alleged that Moonshot used knowledge distillation techniques to reverse-engineer Anthropic's Fable model for its Kimi K3 product.
Read on TechCrunch AI →Nearly 200 major US utility companies and data center operators have signed President Trump's "rate payer protection pledge" to prevent AI infrastructure costs from increasing consumer electricity bills. The commitment, announced in March, addresses public concerns about rising energy consumption from AI development.
Read on The Verge AI →A federal judge approved Anthropic's $1.5 billion settlement with authors who claimed their copyrighted books were used to train AI models without permission. Authors will receive approximately $3,000 per book, making it the largest copyright recovery settlement on record.
Read on The Verge AI →U.S. Treasury Secretary Scott Bessent indicated the administration may impose sanctions on Chinese open-source AI models for intellectual property theft, extending efforts to restrict China's AI development.
Read on TechCrunch AI →Researchers analyzed frameworks for international agreements that would limit advanced AI development, examining how and when such restrictions could be relaxed. They recommend establishing a time-bound agreement with a dedicated organization to assess safety conditions before lifting limitations, with provisions for emergency withdrawal.
Read on arXiv cs.AI →Anthropic has received final approval for a $1.5 billion copyright settlement. While this resolves one legal case, broader questions remain about whether using copyrighted content for AI training is permissible.
Read on TechCrunch AI →David Sacks has resigned from his role as director of the Center for AI Standards and Innovation (CAISI), continuing a pattern of turnover in the position. The leadership vacancy raises questions about stability in the organization's direction and operations.
Read on TechCrunch AI →Sony Music Entertainment filed a lawsuit against AI music generator Udio, alleging copyright infringement of over 30,000 songs including works by Elvis Presley, Beyoncé, and Harry Styles. The filing claims the listed songs represent only a portion of Udio's infringing content, continuing Sony's legal action initiated in 2024 against multiple AI music platforms.
Read on The Verge AI →OpenAI and U.S. policymakers are concerned about open-weight language models from China, raising questions about how to regulate AI while maintaining commercial viability. The debate reflects tensions between national security interests and the business model challenges of controlling AI technology.
Read on TechCrunch AI →YouTube updated its monetization policies to provide clearer guidelines on which AI-generated and low-quality videos are ineligible for ad revenue. The changes aim to better define what content violates the platform's standards for creator earnings.
Read on TechCrunch AI →A multinational survey across seven countries found citizens strongly favor AI safety regulations, public governance over industry self-regulation, and international regulatory frameworks. The study reveals gaps between public preferences for stringent oversight and current dominant regulatory approaches prioritizing innovation and private sector involvement.
Read on arXiv cs.AI →Apple filed a trade secrets lawsuit against OpenAI, claiming the company hired over 400 former Apple employees and alleging misconduct by OpenAI's hardware leadership. The legal action could complicate OpenAI's reported plans to pursue an initial public offering.
Read on TechCrunch AI →San Francisco's City Attorney sent cease-and-desist letters to Apple and Google demanding removal of 13 face-swap applications from their app stores, citing that these tools are primarily used to create non-consensual intimate imagery targeting women and girls.
Read on Wired AI →Anthropic has backed AI transparency legislation in California and New York, though the company's state policy leader now suggests these laws may already be inadequate. The firm is advocating for faster regulatory updates to keep pace with AI development.
Read on Wired AI →The European Union has ordered Google to grant rival AI assistants and search engines greater access to Android and Google Search under its digital antitrust regulations. The rulings could significantly reduce Google's control over these platforms and create new competitive opportunities for competitors.
Read on The Verge AI →A research paper analyzes two AI governance models: one centralizing authority with frontier model providers and another distributing it to organizations deploying AI systems. Examining major regulatory frameworks, the authors argue that while upstream provider controls suit capability management, operational authority over high-impact actions should rest with deploying organizations that bear legal and commercial consequences.
Read on arXiv cs.AI →Vint Cerf, co-creator of TCP/IP, is developing a standard to identify AI agents operating on the open internet. The initiative aims to establish protocols for recognizing and potentially managing autonomous AI systems as they become more prevalent online.
Read on TechCrunch AI →OpenAI has proposed a 'reverse federalism' governance model for AI, where state-level regulations work together to establish a cohesive national framework for safe and democratic AI development. The approach suggests states can pioneer regulations that inform broader federal standards.
Read on OpenAI →Researchers propose domain-specific standardization recommendations for algorithmic hiring systems to comply with the EU AI Act's high-risk requirements. The framework addresses risk management, data quality, transparency, and fairness across the recruitment system lifecycle, offering practical guidance beyond existing horizontal AI governance approaches.
Read on arXiv cs.AI →AAAI-26 organizers discovered 141 dual submissions—papers simultaneously submitted to multiple venues without disclosure—using automated similarity detection and manual review. The conference attributes the growing problem partly to generative AI tools making paraphrased duplicates harder to identify, and recommends policy updates and cross-venue coordination to address it.
Read on arXiv cs.AI →Major publishers including Hachette, Cengage, and Elsevier filed a lawsuit claiming Google used their copyrighted works to train AI models without obtaining proper licenses or permissions.
Read on TechCrunch AI →DeepMind's CEO Demis Hassabis has proposed creating an independent standards organization similar to FINRA to evaluate advanced AI models and establish guidelines for their deployment.
Read on TechCrunch AI →Former Meta employees filed a lawsuit claiming the company used biased AI systems to identify workers for layoffs, with the algorithm allegedly penalizing employees on parental or medical leave by failing to account for their protected status. The lawsuit alleges the AI-driven process disproportionately targeted workers exercising legal leave protections.
Read on The Verge AI →New York Governor Kathy Hochul announced a temporary pause on approving new large data centers, citing concerns about rising electricity costs, water depletion, and local autonomy amid the rapid expansion driven by AI infrastructure demands.
Read on TechCrunch AI →Demis Hassabis, CEO of Google DeepMind, proposed establishing a global AI regulatory body with authority to review frontier models before release. He suggested the US should lead this effort due to its technical and economic position, with the organization modeled after existing financial regulators.
Read on The Verge AI →New York Governor Kathy Hochul signed the nation's first statewide data center moratorium, blocking new environmental permits for facilities exceeding 50 megawatts for up to one year. The pause aims to allow the state to develop regulations addressing energy costs and environmental concerns before large-scale data centers can expand operations.
Read on The Verge AI →The Department of Housing and Urban Development declined to release documents detailing how Elon Musk's DOGE initiative used artificial intelligence in housing policy decisions, citing a legal privilege that doesn't actually exist. The agency's refusal to disclose details about the AI application raises questions about government transparency and accountability.
Read on Wired AI →Apple has filed a lawsuit against OpenAI, claiming that former Apple employees now working at the company misappropriated trade secrets. The suit names OpenAI's chief hardware officer and other ex-Apple staff, along with IO Products (Jony Ive's hardware startup acquired by OpenAI), as defendants in the alleged theft.
Read on The Verge AI →Apple filed a lawsuit against OpenAI, alleging the company encouraged former Apple employees to transfer confidential information including internal presentations, hardware prototypes, and supplier details. The case centers on allegations of trade secret theft through employee recruitment.
Read on Wired AI →News publishers have filed a motion seeking sanctions against OpenAI, alleging the company withheld tools and datasets that could demonstrate whether ChatGPT outputs contain copyrighted news material. The move represents an escalation in the ongoing copyright lawsuit between major publishers and OpenAI.
Read on TechCrunch AI →Google is expanding disclosure requirements for AI-generated content in advertisements beyond just election ads. Advertisers will now need to reveal when ads use AI to create synthetic or digitally altered content, addressing concerns about misleading AI-generated material.
Read on TechCrunch AI →A TechCrunch report questions the transparency of government review processes for advanced AI models, noting that the specific discussions between U.S. authorities and companies like OpenAI and Anthropic regarding frontier model safety assessments remain undisclosed.
Read on TechCrunch AI →OpenAI has launched a bug bounty program focused on biological risks related to AI systems. The initiative invites security researchers to identify and report vulnerabilities that could enable misuse of AI for biological threats.
Read on OpenAI →OpenAI outlined its framework for collaborating with government and national security entities, emphasizing principles around responsible AI deployment, maintaining democratic oversight, and protecting public safety in these partnerships.
Read on OpenAI →Meta has enabled its Muse Image model to use publicly shared Instagram photos for generating AI-created images, requiring users to actively opt out rather than opt in for this feature. The default setting allows Meta to incorporate user content into AI training and generation unless individuals take steps to disable the practice.
Read on Wired AI →Google updated its privacy settings to enable AI training on user data by default. Users who prefer not to contribute their information to Google's AI models can disable this setting through their account preferences.
Read on TechCrunch AI →Universidad Politécnica de Madrid is developing a comprehensive framework for generative AI adoption in higher education that prioritizes pedagogical redesign and AI literacy over detection and punishment. Rather than focusing on plagiarism detection, the university advocates for clear usage policies, authentic assessment methods, and critical thinking development to treat students as informed participants in AI use.
Read on arXiv cs.AI →Cloudflare is implementing a policy requiring AI companies to distinguish between search engine crawlers and AI training crawlers by September 15, with non-compliant traffic set to be blocked by default across many publisher websites. This move aims to ensure AI companies properly compensate publishers for content used in training.
Read on TechCrunch AI →The Trump administration lifted export restrictions on Anthropic's Fable 5 and Mythos 5 AI models after the company implemented new security measures. The removal of restrictions came with unspecified conditions attached.
Read on Wired AI →The Trump administration removed restrictions that had been placed on Anthropic's Mythos and Fable AI models. Anthropic announced it would restore public access to the Fable model starting July 1.
Read on TechCrunch AI →Anthropic received approval from the Department of Commerce to restore access to Claude Fable 5 and Mythos 5 models after export controls were lifted. The company plans to resume offering these consumer-facing models following weeks of negotiations with the Trump administration.
Read on The Verge AI →The Trump administration is removing export restrictions on Anthropic's Mythos and Fable AI models, reversing course after previously ordering the company to restrict access for foreign nationals. The policy shift affects how widely these advanced models can be distributed internationally.
Read on Wired AI →Researchers analyzed US and EU policy documents on AI content watermarking and detection, identifying significant gaps between what regulations propose and what current technology can actually accomplish. The study reveals misalignments in how policymakers frame content tracking requirements versus technical feasibility.
Read on arXiv cs.AI →TIDAL implemented a policy to block AI-generated music from generating revenue on its platform. The move reflects streaming services' growing efforts to protect human artists from unpaid AI-created content.
Read on TechCrunch AI →U.S. lawmakers plan to introduce an updated Health and Location Data Protection Act that would prohibit the sale of Americans' health and location data, including information disclosed to AI chatbots. The expanded bill aims to prevent both data brokers and other companies from trading this sensitive information in an era where AI systems increasingly collect personal health details.
Read on The Verge AI →Prosecutors in an arson trial for the Palisades fire used ChatGPT logs as evidence, including records of the defendant generating fire images, asking about anger management, and discussing wealth inequality. The logs were presented alongside traditional evidence like location data and witness testimony to support charges against the defendant.
Read on The Verge AI →Anthropic's Mythos 5 model has received partial approval from the Trump administration following two weeks of negotiations, becoming available to select organizations. However, the public version called Fable 5 remains unclear regarding deployment timelines.
Read on The Verge AI →The Trump administration approved Anthropic's request to provide access to its advanced Mythos AI model to a limited set of US organizations, including companies and government agencies, following extended negotiations with the White House.
Read on Wired AI →OpenAI delayed the rollout of GPT-5.6 following a government request, stating that such access restrictions should not become standard practice. The company argued that limiting availability prevents users, developers, and security professionals from accessing beneficial AI tools.
Read on TechCrunch AI →The White House requested OpenAI postpone releasing its GPT-5.6 models, following a similar incident where Anthropic removed its most advanced AI models from service. The reasons for these delays remain unclear.
Read on Wired AI →Anthropic has kept its advanced Mythos-class models offline for two weeks following a Trump administration directive, with company executives engaging in ongoing Washington negotiations. The company has provided no updates on resolution prospects or timeline for restoring the models' availability.
Read on The Verge AI →The Trump administration has requested that OpenAI delay its full public release of GPT-5.6, instead distributing it initially to a limited set of partners. The request stems from safety considerations regarding the model's capabilities.
Read on TechCrunch AI →The Trump administration requested OpenAI delay and stagger the release of GPT-5.6 due to security concerns. OpenAI will initially offer the model only to select enterprise customers in a limited preview, with the federal government approving individual access requests during this period.
Read on The Verge AI →European chip equipment makers face pressure from U.S. restrictions on advanced technology sales, with the MATCH Act proposing to limit access to older deep ultraviolet manufacturing tools that were already commercially available a decade ago.
Read on TechCrunch AI →Google now stores media uploads from Search interactions, including images used in reverse image searches, to train its AI models. Users can opt out of this data collection through their account settings.
Read on Wired AI →A U.S. congresswoman clarified that her staff used AI only for proofreading in a defense bill amendment summary, not for drafting legislation itself, after social media posts showed evidence of Claude AI being used in the document.
Read on The Verge AI →Corporate-backed super PACs invested $27 million in a local election, highlighting significant spending by tech and business interests in what is typically a grassroots-level political race. The intervention raises questions about corporate influence on local governance and political processes.
Read on The Verge AI →OpenAI is contributing to the development of shared standards for advanced AI systems, focusing on evaluation methods, safety practices, and international collaboration via the Appia Foundation.
Read on OpenAI →TechCrunch's Equity podcast explored the Trump administration's recent actions against Anthropic and their potential implications for the broader AI industry. The discussion examined the motivations behind these moves and how they could reshape the competitive landscape.
Read on TechCrunch AI →A TechCrunch analysis argues that export controls on cybersecurity technology, attempted for three decades, have consistently failed to prevent distribution. The piece questions whether restricting Anthropic's Mythos security model would succeed where previous efforts have not.
Read on TechCrunch AI →The US government required Anthropic to withdraw two recently released models over security vulnerabilities discovered by Amazon researchers, but cybersecurity experts disputed the decision, arguing similar issues exist in competing systems. The incident has generated debate about whether the regulatory action inadvertently benefits Anthropic's public perception.
Read on TechCrunch AI →The US government forced Anthropic to withdraw its Fable 5 and Mythos 5 models citing national security concerns after researchers discovered guardrail vulnerabilities. Cybersecurity experts and Anthropic disputed the ban, arguing similar vulnerabilities exist in competing models.
Read on TechCrunch AI →The U.S. government alleged that ASML's advanced chip manufacturing equipment reached China, but ASML denied the claim. The disagreement highlights tensions over semiconductor technology exports and export compliance.
Read on TechCrunch AI →Researchers argue that standardized data formats are essential infrastructure for advancing humanoid robotics, enabling robots to learn from accumulated physical experience across different systems and organizations. The authors, developing ISO standards for humanoid robot datasets, identify data silos and inconsistent evaluation as key bottlenecks, proposing comprehensive standards that preserve embodied context like timing, calibration, and task relationships.
Read on arXiv cs.AI →Anthropic faces restrictions on distributing Claude Mythos and Fable 5 models following Trump administration actions, though the specific violations remain unclear. The situation highlights ambiguity in how AI regulations are being applied and enforced at the federal level.
Read on Wired AI →The Federal Energy Regulatory Commission issued an order requiring grid operators to expedite interconnection processes for artificial intelligence data centers. However, the directive does not address underlying electricity supply constraints that could limit the ability to actually power these facilities.
Read on TechCrunch AI →A political action committee called Guardrails, backed by tech workers, launched a $5 million campaign funded through small donations from industry employees. The effort aims to counter major tech companies' significantly larger political spending on AI-related policy issues.
Read on TechCrunch AI →The US government imposed export controls on Anthropic's Fable 5 AI model and its underlying Mythos architecture shortly after public release, restricting access for foreign nationals including those employed in the US. Anthropic subsequently took both models offline, citing compliance concerns, creating uncertainty about the decision-making process for AI safety restrictions and broader implications for the industry.
Read on The Verge AI →The UK Home Office plans to implement facial recognition for age verification of asylum seekers despite internal testing revealing the technology produces unreliable results. The government is proceeding with deployment despite documented risks of significant errors that could affect immigration outcomes.
Read on Wired AI →The White House directed Anthropic to revoke SK Telecom's access to Claude Mythos, citing alleged connections to China. This action occurred shortly before Anthropic took its most advanced AI models offline.
Read on Wired AI →French and Indian leaders expressed concerns at the G7 summit that the U.S. could unilaterally restrict access to American AI systems, with recent industry disruptions highlighting the vulnerability of relying on U.S.-controlled technology infrastructure.
Read on TechCrunch AI →Anthropic was ordered by the Trump administration to restrict access to its latest AI models (Fable 5 and Mythos 5) to all foreign nationals and non-US employees, citing national security export controls. The administration did not publicly disclose the legal justification, marking what appears to be the first application of export rules to limit AI model access.
Read on The Verge AI →Trump administration officials have demanded that Anthropic implement unbypassable safeguards in Claude 5 as a condition for rerelease, but cybersecurity experts contend that completely eliminating jailbreak vulnerabilities is technically infeasible for large language models.
Read on Wired AI →The Verge reports on internal tensions between the White House and Anthropic over an AI project called Fable. The article draws parallels to factional conflicts within Trump's first administration, suggesting different government interests are competing over AI policy decisions.
Read on The Verge AI →Recent US government restrictions on advanced Anthropic models highlight an emerging challenge: AI systems with sophisticated hacking abilities are becoming increasingly difficult to contain and will likely proliferate regardless of regulatory efforts.
Read on Wired AI →The Department of Justice argued that xAI's use of unpermitted gas turbines is necessary for national security, with the Pentagon reportedly depending on the power infrastructure for operations. The statement suggests the DOJ is defending xAI's continued operation of the turbines despite regulatory violations.
Read on TechCrunch AI →The Justice Department argued in court filings that xAI should be exempt from an NAACP environmental lawsuit, asserting the company plays a critical role in national security and military operations. The DOJ's position emphasizes xAI's importance to defense infrastructure despite environmental concerns raised by the plaintiffs.
Read on Wired AI →Anthropic received a US export control directive requiring it to suspend access to its Mythos 5 and Fable 5 models to foreign nationals, including foreign employees. The company disabled the products and sought to negotiate with the Trump administration to reverse the order.
Read on The Verge AI →Anthropic executives met with White House officials to discuss concerns about Claude Fable 5's potential risks, but the two parties remain unable to reach agreement on the severity of those risks following their conversations.
Read on Wired AI →The Trump administration compelled Anthropic to withdraw recently developed cybersecurity models, signaling government willingness to intervene in AI development. The decision appears motivated by political rather than technical security concerns.
Read on TechCrunch AI →The White House ordered Anthropic to block foreign access to its newly released Fable 5 and Mythos 5 models after researchers discovered potential security vulnerabilities that could enable cyberattacks. Anthropic complied by removing access for all users, though the company disagreed with the decision, arguing that a narrow jailbreak vulnerability shouldn't warrant recall of models deployed at scale.
Read on The Verge AI →The Trump administration ordered Anthropic to take its newest AI models offline and block access for foreign nationals, including the company's international employees. The incident highlights US government control over frontier AI technology and has prompted discussions about developing non-American AI alternatives.
Read on The Verge AI →Technology industry lobbyists are making a final push for federal AI preemption legislation that would establish uniform national rules and override state-by-state regulations. The effort faces significant political obstacles, including potential Democratic control of Congress after midterm elections and accumulated public opposition.
Read on The Verge AI →Cybersecurity experts are calling on the White House to lift export restrictions on Anthropic's advanced models, contending the limitations will hinder their ability to defend against threats and secure software.
Read on TechCrunch AI →Researchers analyzed contribution policies across six major open-source organizations to address governance challenges posed by AI agents submitting code with minimal human oversight. The study maps existing policies against emerging AI regulations (EU AI Act, NIST AI RMF, ISO standards) and identifies gaps, proposing a framework to harmonize open-source governance with broader AI regulatory requirements.
Read on arXiv cs.AI →A report claims the White House imposed export restrictions on Anthropic's advanced AI models due to concerns that a China-linked group accessed them, citing potential national security risks. The White House has not confirmed the report, and the restriction's exact rationale remains unclear.
Read on The Verge AI →Amazon security research demonstrating potential cybersecurity vulnerabilities in Anthropic's Fable 5 model reportedly influenced White House policy discussions, leading to export control directives that restricted foreign access to the AI system. CEO Andy Jassy shared the findings with government officials before Anthropic implemented the access limitations.
Read on The Verge AI →State attorneys general have launched an investigation into OpenAI, examining various aspects of the company's operations including advertising practices and health data management. The investigation's scope and specific states involved have not been fully disclosed.
Read on TechCrunch AI →Anthropic has disabled access to its Fable 5 and Mythos 5 models following a government order citing national security concerns. The company complied with the directive despite stating the government did not provide detailed justification for the decision.
Read on The Verge AI →A German court determined that Google bears legal responsibility for inaccurate information produced by its AI Overviews feature. The decision establishes that companies operating and managing AI systems are accountable for the harm caused by their outputs.
Read on Wired AI →Anthropic disputed a government decision to halt deployment of its most advanced AI model, arguing that a identified vulnerability doesn't warrant recalling software already in widespread use. The company expressed disagreement with the regulatory action in an official statement.
Read on TechCrunch AI →Anthropic is taking Claude Fable 5 offline following a US government order after authorities identified a method to jailbreak the model. The company disclosed this action in a blog post without providing additional details about the specific vulnerability or regulatory requirements.
Read on Wired AI →
Hi — I'm Scott. I'm building AI Info Forge because the Claude skills market is full of prompts that worked last month and break this month. Verified, drift-tested skills is the version of this market I wish existed. If you've felt the same friction, get on the list and I'll send you early access.
— Scott Sokolowski, Founder