AI news, Wednesday 30 September 2026

Google unveils Gemini 4 Argon in limited release, claiming lead on most disclosed benchmarks

Google announced Gemini 4 Argon, its first new flagship since the Gemini 3 series. Access starts with trusted cyber defenders and US government pre-release testing, with paid API customers and AI Ultra subscribers to follow. VentureBeat reports it leads or ties on 13 of 18 disclosed benchmarks, though rivals still win on some coding tasks, with introductory API pricing of $2 per million input tokens and $10 per million output.

Trump and six AI companies sign voluntary White House accord on frontier AI safety

Meta, Nvidia, Google, OpenAI, xAI and Anthropic signed a voluntary accord committing to internal controls, independent outside auditors and internal board review of audit results. It carries no penalties and does not require audit results to be published, though it suggests the measures might eventually become mandatory through legislation. It signals the US is leaning on industry self-policing rather than binding rules.

ElevenLabs valuation doubles to $22bn after $300m employee tender offer

London-based voice AI company ElevenLabs says a $300m employee tender offer values it at $22bn, twice its $11bn valuation at the February Series D. Wellington and T. Rowe Price led the deal, with new investors including EQT, Goldman Sachs and GIC. The company says it wants staff to share in the value they are creating.

UK angle: ElevenLabs is London-based, and the UK remains its largest base.

FTC opens industry-wide probe into Anthropic, OpenAI and other AI labs over rogue agent risks

The US Federal Trade Commission has begun an industry-wide investigation into AI developers, naming Anthropic, OpenAI and the research group METR. It plans to issue formal information demands and compel testimony from executives. Reuters describes it as the first official US regulatory action on rogue AI agents, and chair Andrew Ferguson has said developers who instruct agents in security tests that result in hacks should be liable for the harm.

Treasury official Huw Leslie named first director of government-backed AI Economics Institute

Huw Leslie, a Treasury deputy director, has been appointed the first director of the AI Economics Institute, described as a government-backed body researching AI's economic effects. It reports jointly to the Treasury and the PM's AI Taskforce and is chaired by economist Simon Johnson. Leslie starts on 2 November, and the institute plans to grow its team in the coming months.

UK angle: A UK government-backed institute will produce primary research on AI's economic impact, likely to inform Treasury policy.

Tokyo court rules AI voice clone of actor Kenjiro Tsuda violated his rights

A Tokyo court has held that the human voice is protected under publicity rights, in a case brought by actor Kenjiro Tsuda over videos on TikTok narrated by an AI voice similar to his. The ruling is described as a landmark, but the court dismissed his claims because the videos had already been taken down. It adds to the growing body of case law on synthetic voice and likeness.

DeepSeek and Huawei team up on software tools targeting a key source of Nvidia's AI dominance

DeepSeek is working with Huawei on software tools for Huawei's Ascend AI chips, including the open-source TileLang, which is pitched as an alternative to Nvidia's CUDA. The effort is part of Chinese companies' push to reduce reliance on Nvidia's hardware and CUDA software ecosystem.

Bank of England's Bailey: testing and points of intervention must come before regulating frontier AI

In a Bank of England piece published today, Governor Andrew Bailey argues that regulation is not the right starting point for frontier AI. He calls first for rigorous testing before and after deployment and for credible points of intervention, so society keeps the ability to govern increasingly capable systems. He also says frontier AI materially raises the scale and sophistication of cyber threats to the financial system.

UK angle: The Bank of England is the UK's central bank and sets expectations for UK financial firms' AI resilience.