Wiredframe Radar logo

Category intelligence

AI News Briefing — April 10, 2026

22 current items analyzed and ranked.

Executive synthesis

AI News Summary

Anthropic dominates this cycle with the release of Claude Mythos, a frontier model so capable at finding cybersecurity vulnerabilities that it's being withheld from public access. Through Project Glasswing, the company is giving partners including Microsoft, Apple, and Google up to $100M in credits to patch thousands of discovered bugs. The 244-page system card also reveals unprecedented AI welfare explorations, including psychiatric evaluations of the model.

Meta made two major moves:

Policy and geopolitics shaped the week heavily:

  • A federal appeals court refused to block the Trump administration's blacklisting of Anthropic, with oral arguments set for May 19
  • OpenAI shelved Stargate UK, citing energy costs and regulation, undermining Britain's £31B AI investment deal
  • A Pentagon AI official profited up to $24M from xAI stock amid government contracts with the company
  • The first conviction under the Take It Down Act was secured against an Ohio man who used AI to create nonconsensual explicit images

Key Themes

Frontier Model Releases · 4AI Policy & Government · 5AI Infrastructure & Investment · 2AI Safety & Responsible Deployment · 3Agentic AI & Enterprise · 5AI Misuse & Deepfakes · 4

Primary evidence

Top Ranked Signals

News Ars Technica - All content Apr 9

AI on the couch: Anthropic gives Claude 20 hours of psychiatry

By Nate Anderson

92 score
AI Analysis

First discussed on LessWrong, the Mythos system card is now getting mainstream attention, Anthropic released a 244-page system card for Claude Mythos, its most capable frontier model, which it decided not to make generally available due to its cybersecurity capabilities. The company also explored whether Claude may have some form of experience or welfare, reportedly subjecting it to 20 hours of psychiatric evaluation.

The AI company Anthropic released a 244-page "system card" (PDF) this week describing its newest model, Claude Mythos. The model is "our most capable frontier model to date," the company says, and supposedly is so good that Anthropic has decided "not to make it generally available." (The company claims that Mythos is too good at finding unknown cybersecurity bugs, and so the model is only being released to select companies like Microsoft and Apple for now.) Whatever the truth of this claim, the
frontier_model_releaseai_safetyai_consciousnessanthropic
News AI News Apr 9

Anthropic keeps new AI model private after it finds thousands of external vulnerabilities

By Dashveenjit Kaur

91 score
AI Analysis

As reported in News yesterday, Anthropic's Claude Mythos has discovered thousands of cybersecurity vulnerabilities across major operating systems and browsers. The company launched Project Glasswing, giving select partners like AWS, Apple, Microsoft, Google, and Nvidia access with up to $100M in usage credits to fix critical infrastructure bugs.

Anthropic’s most capable AI model has already found thousands of AI cybersecurity vulnerabilities across every major operating system and web browser. The company’s response was not to release it, but to quietly hand it to the organisations responsible for keeping the internet running. That model is Claude Mythos Preview, and the initiative is called Project Glasswing. The launch partners include Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganCha
frontier_model_releasecybersecurityai_safetyanthropicresponsible_deployment
News AI (artificial intelligence) | The Guardian Apr 9

Meta debuts new AI model in first test of costly ‘superintelligence’ team

By Reuters

88 score
AI Analysis

Continuing our coverage from yesterday, Meta unveiled Muse Spark, the first AI model from its costly superintelligence team led by former Scale AI CEO Alex Wang, hired in a $14.3B deal. The model is competitive with OpenAI, Google, and Anthropic in language tasks but lags in coding.

Muse Spark was competitive with models from OpenAI, Google and Anthropic in language, but lagged in codingSign up for the Breaking News US email to get newsletter alerts in your inboxMeta on Wednesday unveiled Muse Spark, the first artificial intelligence model from a costly team it assembled last year to catch up with rivals in the AI race.US tech companies are under pressure to prove their huge AI outlays will pay off. The stakes are especially high for Meta after it hired Alex Wang, Scale AI
frontier_model_releasemetaai_competitionmultimodal_ai
News Ars Technica - All content Apr 9

Trump-appointed judges refuse to block Trump blacklisting of Anthropic AI tech

By Jon Brodkin

85 score
AI Analysis

Continuing our coverage from yesterday, A federal appeals court with Trump-appointed judges refused to block the Trump administration's blacklisting of Anthropic, denying its emergency stay motion. Oral arguments are expedited for May 19, with Anthropic having a second related case pending.

A federal appeals court refused to halt the Trump administration's efforts to blacklist Anthropic yesterday, denying the company's emergency motion for a stay. But the court granted the US-based AI firm's request to expedite the case and will hold oral arguments on May 19. The ruling by the US Court of Appeals for the District of Columbia Circuit was issued by a panel of three judges appointed by Republicans, including Trump appointees Gregory Katsas and Neomi Rao. Katsas previously served as de
ai_policyanthropicgovernment_regulationtrump_administration
News aibusiness Apr 9

Meta, CoreWeave In $21B Deal to Expand AI Partnership

By Scarlett Evans

82 score
AI Analysis

Meta and CoreWeave have entered a $21 billion deal to expand their AI infrastructure partnership, representing one of the largest AI compute deals to date.

ai_infrastructuremetafunding_dealscompute
News AI (artificial intelligence) | The Guardian Apr 9

OpenAI shelves Stargate UK in blow to Britain’s AI ambitions

By Aisha Down and Alexandra Topping

80 score
AI Analysis

OpenAI has shelved its Stargate UK data center project, citing high energy costs and regulation. The project was part of a £31B US-UK AI investment deal announced last September, dealing a blow to Britain's AI strategy.

Artificial intelligence company cites high energy costs and regulation for putting landmark project on holdOpenAI has put on hold plans for a landmark UK investment citing high energy costs and regulation, in a blow to the government which has put AI at the centre of its growth strategy.Stargate UK was a part of the UK-US AI deal announced last September, in which US companies appeared to commit £31bn to the UK’s tech sector, part of a larger series of investments intended to “mainline AI” into
ai_infrastructureopenaigeopoliticsenergyregulation
News MarkTechPost Apr 9

Meta Superintelligence Lab Releases Muse Spark: A Multimodal Reasoning Model With Thought Compression and Parallel Agents

By Asif Razzaq

78 score
AI Analysis

Building on yesterday's News coverage of the Muse Spark launch, Technical deep-dive into Meta's Muse Spark reveals it is a natively multimodal reasoning model with thought compression, parallel agents, and visual chain-of-thought capabilities—trained from the ground up for integrated vision-language reasoning rather than bolting modules together.

Meta Superintelligence Labs recently made a significant move by unveiling ‘Muse Spark’ — the first model in the Muse family. Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration. ai.meta.com/static-resource/muse-spark-ev... What ‘Natively Multimodal’ Actually Means When Meta describes Muse Spark as ‘natively multimodal,’ it means the model was trained
frontier_model_releasemetamultimodal_aimodel_architecture
News AI (artificial intelligence) | The Guardian Apr 9

US defense official overseeing AI reaped millions selling xAI stock after Pentagon entered agreement with company

By Aram Roston in Washington

72 score
AI Analysis

Pentagon Under Secretary Emil Michael, who oversees AI efforts, profited up to $24M selling xAI stock after the DOD entered an agreement with Musk's company. Ethics experts say federal law bars officials from actions benefiting their own financial interests.

Expert said federal law bars officials from taking actions in their jobs that benefit their own financial interestsA high-profile US defense department official who oversees the agency’s artificial intelligence efforts made a profit of up to $24m selling a private investment he held in Elon Musk’s AI company earlier this year, according to government ethics records released this month. The value of his stake totaled a maximum of a million dollars when he joined the department.Emil Michael, who i
ai_policyethicsgovernmentxaitrump_administration
News Ars Technica - All content Apr 9

First man convicted under Take It Down Act kept making AI nudes after arrest

By Ashley Belanger

62 score
AI Analysis

Continuing our coverage from yesterday, An Ohio man became the first person convicted under the Take It Down Act for creating and distributing AI-generated explicit images of at least 10 victims. He used over 24 AI platforms and 100 web-based models to create hundreds of nonconsensual images.

An Ohio man became the first person convicted under the Take It Down Act after pleading guilty to creating and sharing both real and AI-generated explicit images of at least 10 victims without their consent. According to a Justice Department press release, 37-year-old James Strahler II used AI tools to create fake sexualized images to harass at least six women he knew. In some images, he depicted one victim engaged in sex with her father and shared that image with her mother and co-workers. He a
ai_policydeepfakeslegal_precedentai_misuse
News aibusiness Apr 9

New Anthropic Tool Speeds up AI Agent Development for Enterprises

By Graham Hope

60 score
AI Analysis

As reported in News yesterday, Anthropic released a new enterprise tool to accelerate AI agent development, part of a flurry of releases as the company competes intensely with OpenAI in the agentic AI space.

The product comes amid a flurry of releases from the generative AI vendor, which is locked in a tight battle with rival OpenAI.
agentic_aianthropicenterprise_aiproduct_launch
News AI News Apr 9

Agentic AI’s governance challenges under the EU AI Act in 2026

By Joe Green

58 score
AI Analysis

As EU AI Act enforcement begins in August 2026, agentic AI systems pose unique governance challenges because they can act autonomously without clear audit trails. IT leaders face substantial penalties for governance failures in high-risk areas.

AI agents hold the promise of automatically moving data between systems and triggering decisions, but in some cases, they can act without a clear record of what, when, and why they undertook their tasks. That has the potential to create a governance problem, for which IT leaders are ultimately responsible. If an organisation can’t trace an agent’s actions and don’t have proper control over its authority, leaders can’t prove that a system is operating safely or even la
regulationagentic_aieu_ai_actgovernance
News Feed: Artificial Intelligence Latest Apr 9

The 70-Person AI Image Startup Taking on Silicon Valley's Giants

By Maxwell Zeff

55 score
AI Analysis

Black Forest Labs, a 70-person startup, continues to compete with major tech companies in AI image generation and is now expanding into physical AI applications.

Black Forest Labs has long punched above its weight in the AI image generation space. Its next move? Powering physical AI.
image_generationstartupsai_competition