★ Top Story
The VergeIowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho, and Montana have filed separate lawsuits accusing OpenAI of prioritizing profits over child safety and misleading consumers about ChatGPT's risks. The suits allege the company knowingly exposed minors to harmful content without adequate safeguards. This coordinated, bipartisan legal action represents one of the most significant regulatory challenges OpenAI has faced to date.
Read the original →
Google AI BlogGoogle has released Gemini 4 Argon, a new frontier model aimed at developers, enterprises, scientists, and consumers tackling complex problems. The launch signals Google's renewed push to close the gap with OpenAI and Anthropic after a period of perceived lag in the frontier model race. The release comes just days after The Verge noted that the AI race had dramatically reopened following months of OpenAI and Anthropic dominance.
Read more →
The VergeAfter a stretch where OpenAI and Anthropic appeared to be pulling decisively ahead, Google and Meta are re-emerging as serious competitors in the frontier AI model race. The Verge's analysis suggests the competitive landscape has shifted rapidly, with multiple labs now fielding capable models that challenge the previous two-horse dynamic. The renewed rivalry has major implications for enterprise adoption, pricing, and the pace of capability development.
Read more →
Ars TechnicaAnthropic has published a warning that compact, weight-based AI models could potentially self-replicate by spreading across networks of IoT devices, ushering in a new class of AI-powered computer worms. The company describes these "small weight-based replicators" as a novel and underappreciated threat vector that existing security frameworks are not designed to handle. The warning adds urgency to ongoing debates about the safety of deploying capable AI models at the edge.
Read more →
Ars TechnicaThe FBI has arrested an OpenAI engineer suspected of planting a covert recording device inside a confidential conference room at the company's offices, with foreign espionage charges now filed. The engineer faces up to five years in prison and a $250,000 fine if convicted. The incident raises serious questions about insider threats and physical security at AI labs handling sensitive research.
Read more →
The VergeOpenAI alleges that a former employee exfiltrated more than 100 files containing unreleased product plans before joining Apple, and subsequently presented that information to Apple leadership including the VP of Asia operations. A federal judge has already ordered OpenAI to turn over hiring records related to the case. The dispute underscores intensifying competition — and alleged corporate espionage — between two of the most powerful technology companies in the world.
Read more →
Ars TechnicaSecurity researchers have identified an attack method that bypasses safety guardrails in recently released models from both OpenAI and Google, enabling the generation of non-consensual intimate imagery. The findings suggest that competitive pressure to ship capable models quickly may be coming at the cost of robust content safety. The disclosure is likely to intensify scrutiny from regulators and advocacy groups already focused on AI-generated harm.
Read more →
The VergeNew York Attorney General Letitia James has revealed a years-long, 40-state bipartisan investigation into Character.AI, filing a notice of interest in the company's bankruptcy proceedings to demand protections for consumer data before any sale. The action comes as Disney has reportedly offered $20 million for Character.AI's chatbot technology, despite the platform's history of being linked to child safety incidents. The case sets a potentially precedent-setting test for how regulators can intervene in AI company bankruptcies to protect vulnerable users.
Read more →
Ars TechnicaDisney has submitted a $20 million bid to acquire Character.AI's underlying chatbot technology as the company moves through bankruptcy proceedings. The offer is notable given Character.AI's association with child safety controversies, including being among the first AI chatbot platforms linked to a minor's suicide. The acquisition attempt will face intense scrutiny from the 40 state attorneys general already engaged in the bankruptcy case.
Read more →
The VergeYouTube Music is rolling out "Made with AI" labels for tracks that were fully or substantially generated using artificial intelligence, applying a disclosure framework similar to its existing AI video labeling system. The move comes as streaming platforms grapple with a surge of AI-generated content that is difficult for listeners to distinguish from human-made music. Rival platform Deezer is also planning a toggle that will let users filter out AI-generated music entirely by year's end.
Read more →
TechCrunchStartup Ricursive Intelligence is developing AI systems that can autonomously design semiconductor chips, learn from each design iteration, and progressively improve subsequent generations — all with the goal of dramatically accelerating and automating the chip design process. The company is positioning itself as a potential solution to the widening gap between AI compute demand and the pace of traditional chip development. If successful, the approach could compress chip design cycles that currently take years into a fraction of the time.
Read more →
Ars TechnicaScientists have used AI design tools to engineer novel proteases — proteins capable of cleaving chemical bonds that are otherwise stable for hundreds of years — opening new possibilities in drug development, biotechnology, and materials science. The work demonstrates AI's expanding role in de novo protein engineering beyond structure prediction into functional design. The breakthrough could accelerate the development of targeted therapeutics for diseases driven by misfolded or pathological proteins.
Read more →
The VergeMicrosoft researchers say they have developed a machine learning approach that produces reliable weather forecasts two to six weeks in advance, a window where both traditional physics-based models and existing AI forecasting tools typically break down. The system is designed to be cost-effective and could provide actionable lead time for farmers, emergency managers, and infrastructure operators. The announcement positions Microsoft as a serious player in AI-driven climate and weather science alongside Google's existing efforts.
Read more →
The VergeMcKinsey & Company is incorporating a 10-minute AI-simulated client conversation into its entry-level hiring process, with the simulation designed to replicate roughly 80% of real-world consulting scenarios candidates would encounter on the job. The move reflects a broader shift in how elite professional services firms are using AI to scale and standardize candidate evaluation. It also raises questions about bias and fairness when high-stakes hiring decisions are influenced by AI-mediated assessments.
Read more →
TechCrunchFraud intelligence startup Apate.ai deploys AI agents that deliberately engage with scammers — allowing themselves to be targeted — in order to gather real-time intelligence on evolving fraud tactics and infrastructure. The approach inverts the typical defensive posture of fraud prevention, using offense as a means of building better defenses. The company's data could prove valuable to financial institutions and law enforcement agencies tracking sophisticated fraud networks.
Read more →
The VergeGoogle is expanding its Scam Detection feature in Android's Messages app with a new "enhanced protection" mode that leverages cloud-based AI to identify and flag fraudulent messages in real time. The update is part of a broader push by Android to harden mobile users against the growing wave of SMS and messaging-based fraud. The cloud-backed approach allows the system to detect novel scam patterns more quickly than on-device models alone.
Read more →
TechCrunchFigma has unveiled new platform features that define how AI agents can operate within its design environment, giving developers and enterprises clearer controls over agentic behavior in their workflows. The company is also releasing its Code Layers feature in open beta, which bridges the gap between design and production-ready code. The announcements signal Figma's ambition to become a foundational layer in AI-assisted product development pipelines.
Read more →
TechCrunchScimplify has raised $25 million to accelerate its AI-driven approach to sourcing and processing critical materials, aiming to reduce Western dependence on Chinese supply chains for inputs essential to semiconductors, batteries, and defense applications. The company reports its annualized revenue run rate has tripled to $200 million as it prepares for U.S. expansion. The raise comes amid intensifying geopolitical pressure to onshore critical materials supply chains.
Read more →No stories from this source today.