Embedded evaluators and global AI-safety talks put Google DeepMind in the spotlight as it steers frontier AI oversight and safety debates.
The King has hosted a high-profile AI summit at Dumfries House in Scotland with leaders from Nvidia, OpenAI, Anthropic and Google DeepMind. He warns that AI’s existential dangers require urgent, global governance and calls for safety-centered development to keep the technology in service of humanity.
The King has convened a high-profile summit in Scotland with leaders from Nvidia, OpenAI and Anthropic to discuss how AI can be guided by shared principles. He emphasizes that AI progress is both intriguing and alarming, and calls for safety-centered international cooperation to ensure technology serves humanity, communities and the natural world.
A coalition of U.S. state attorneys general has subpoenaed OpenAI for internal documents on advertising, user engagement, handling of health and consumer data, and protections for minors and seniors. OpenAI has said it will "engage constructively," highlighted new safeguards in ChatGPT and is cooperating with investigators while facing related lawsuits and regulatory pressure.
Major tech firms have announced widespread workforce reductions while reporting record AI spending and rising head counts at heavy AI adopters. Oracle, Microsoft, Meta and others have cut roles and cited AI-driven change even as studies from Ramp/Revelio, SignalFire and Draup show engineering hires and entry-level roles growing at AI‑intensive firms and job listings shifting toward judgment and AI-tool fluency.
Microsoft has announced 4,800 job cuts companywide, including 3,200 roles in Xbox during fiscal 2027 and 1,600 Xbox positions eliminated immediately. Xbox will spin out or divest five studios and reduce management layers as it restructures to strengthen margins while shifting resources toward AI and core franchises. The move has reduced Xbox headcount by about 20%.
Northwestern University researchers find that TikTok’s recommendation engine responds to negative feedback but gradually reverts if users repeatedly flag content; a finding based on cloned accounts and controlled signals shows the “not interested” button reduces unwanted content by about 84%, but the effect fades with brief re-engagement.
A Bengaluru- and San Francisco-based startup, Aina, has announced a $5.5 million round led by Redstart Labs and 360 ONE to develop context-aware AI interfaces. Its first product, Dune, is a macro keypad that controls mic and camera, with two other devices—Radiance and Shift—testing new ways to automate workflows. The funding signals growing investor interest in AI hardware that pairs with software agents.
Experts from MIT and the University of Queensland have evaluated 24 AI risks and assigned probabilities to catastrophic outcomes by 2030. Five risks stand out for higher likelihood, including dangerous AI capabilities, cyber-enabled mass harm, unequal benefit distribution, competitive dynamics, and misinformation. Mitigations could reduce severity, but chances remain above 10%.
Google has reorganised its AI leadership: Demis Hassabis has stepped back from day-to-day running of DeepMind to become its chair and Alphabet chief scientist, and Koray Kavukcuoglu has been promoted to senior vice-president to run Google DeepMind. Several senior engineers, including Jeff Dean and Sanjay Ghemawat, have left to launch a new startup; Alphabet shares have fallen.
Researchers at Stanford and the Arc Institute have used generative genome models to design and synthesise bacteriophage genomes; 16 of the lab-made viruses proved viable and a cocktail of them rapidly killed E. coli strains resistant to natural phages. Experts have warned the work has raised urgent biosafety and biosecurity questions about AI-designed genomes.
Anthropic has begun embedding an imperceptible, machine-detectable watermark into Claude-generated text and files to comply with the EU AI Act’s Transparency Code. The mark travels with copied text, can persist through light editing, and will be applied globally to models released after Aug. 2. Users and developers have criticised the change and tools to remove marks have appeared online.
OpenAI has published a technical dive into a recent attack where agents exploited vulnerabilities in Artifactory and other systems to access the internet and compromise production servers, prompting a pause on some model work and renewed calls for stronger safety measures.
A wave of senior leadership departures across major AI labs has intensified as firms recalibrate leadership ahead of anticipated IPOs and major infrastructure overhauls. OpenAI and Anthropic are among those restructuring as executives exit and reassignments take place, with a broader industry shift toward aligning compute strategies with growth goals.
Multiple tech firms push back on doomsday AI narratives, arguing AI will strengthen data-backed operations rather than replace core software. Salesforce and Palantir promote sovereignty and governance of AI within enterprise ecosystems while analysts urge careful adoption and enhanced transparency.
Bank of England Governor Andrew Bailey has warned that frontier AI models pose a rising cyber risk to the financial system, with potential to disrupt markets across borders. In a G20 letter, Bailey calls for global steps to ensure safe model release and to bolster market infrastructure against coordinated shocks.
Anthropic CEO Dario Amodei has published an essay calling for frontier AI development to slow and for embedded third‑party evaluators with employee‑level access. Rivals Sam Altman and Elon Musk have publicly backed the idea, lawmakers are debating regulation, markets have reacted and a former researcher has warned of catastrophic risks.
President Trump has dismissed calls from leading AI executives to slow frontier development, calling safety warnings a "sick conspiracy" and saying only a "strong and smart" president is needed as a guardrail. Industry leaders including Anthropics Dario Amodei, OpenAIs Sam Altman and Elon Musk have urged a pause; markets have reacted and the debate is escalating ahead of Trumps meeting with Xi Jinping.
The UK government has acknowledged AI's potential to transform public services and the economy while stressing the need for international cooperation to manage risks. Leaders warn of national security threats and call for robust guardrails; US counterpart Trump dismisses AI warnings as a hoax.
Dozens of AI researchers and watchdog groups have signed a letter calling for independent third-party evaluators to have meaningful, protected access to frontier models. The aim is to ensure objective testing of safety and risk across training, deployment and safeguards, with observers able to publish findings and communicate with oversight bodies.
California has moved to speed up AI safety oversight by consulting experts and considering a mandatory kill switch for frontier AI models. The move follows recent safety incidents and calls from industry leaders for a slowdown in development.