Thursday, 1 October 2026
The last 24 hours' hottest news in AI, sorted by field, each story with a line on why it matters.
Story of the day
The single most consequential story of the day.

- new valuation
- $22B
- employee tender offer
- $300M
- revenue from enterprise
- 55%
- agent conversations per week
- 15M
ElevenLabs reaches $22 billion valuation after staff tender
ElevenLabs closed a $300 million employee tender offer led by Wellington and T. Rowe Price that values the voice AI company at $22 billion, double its Series D valuation from February. Enterprise customers now make up 55% of its revenue as its voice agents handle more than 15 million conversations a week, up threefold since February.
Why it's the story of the dayIt shows investors betting heavily that voice-based AI agents, not just text chatbots, will become how businesses and governments run customer service.
Models & labs
New models, the labs behind them, and the rules they answer to.
Google launches Gemini 4 Argon but limits access over cyber risks
Google introduced Gemini 4 Argon, its newest frontier model, but is only giving it to selected cybersecurity partners through its Fairwind programme while it keeps testing safeguards. Argon replaces the cancelled Gemini 3.5 Pro and scored 68% on a benchmark for fixing software vulnerabilities, and found a flaw in hospital software that earlier models had missed.
Why it mattersIt shows a top AI lab treating a model's hacking ability as risky enough to control who gets to use it.
Tech Wire Asia ↗Cohere launches Embed 5 for enterprise document search
Cohere released Embed 5, a new family of embedding models for enterprise search and retrieval, including a cheaper Fast tier priced at $0.08 per million tokens. Cohere says Embed 5 Pro scores highest in its tests on financial documents, parsed PDFs and visually complex files.
Why it mattersFaster, cheaper document search helps AI assistants give accurate answers instead of guessing.
Cohere ↗Video generation
The models and tools that make video.
Luma launches Variants to resize ads automatically
Luma's new Variants tool takes one approved static ad and automatically produces versions for five standard ad formats and multiple languages, while keeping the logo, headline and call to action intact. It is live now inside Luma's Discover tab.
Why it mattersIt could save marketing teams the manual work of rebuilding the same ad for every size and market.
Luma AI ↗Runway unveils Praxis-1, an open robot-control model
Runway introduced Praxis-1, its first open-weight 'world action model', which uses its video-generation technology to control real robots. Early partners including Noble Machines, Standard Bots and Ultra are testing it on their own hardware ahead of a public release.
Why it mattersTraining robots on ordinary video instead of costly hands-on demonstrations could make useful robots far cheaper to build.
Runway ↗Creator tools
The apps and platforms people make video with.
ElevenLabs reaches $22 billion valuation after staff tender
ElevenLabs closed a $300 million employee tender offer led by Wellington and T. Rowe Price that values the voice AI company at $22 billion, double its Series D valuation from February. Enterprise clients, including Stripe, Deutsche Telekom and the governments of Ukraine and Greece, now make up 55% of its revenue as its ElevenAgents handle more than 15 million conversations a week.
Why it mattersIt shows investors betting heavily that voice-based AI agents, not just chat, will be how businesses and governments use AI.
ElevenLabs ↗Science & health
AI in biology, medicine, climate and maths.
Google's AI model ranks best for predicting flu hospitalisations
The CDC says a forecasting model built with Google AI was the most accurate of 39 models at predicting US flu hospital admissions during the 2025-26 season. It was developed using Empirical Research Assistance, a Google tool for generating scientific algorithms described in a recent Nature paper.
Why it mattersBetter flu forecasts help hospitals and health officials plan for how many patients to expect each week.
Google ↗Anthropic eases AI limits for vetted biology researchers
Anthropic launched the Life Sciences Verification Program, giving vetted life-science organisations more permissive access to its Mythos, Opus and Sonnet models for drug discovery, clinical development and manufacturing work normally blocked by its safety filters. Applicants are reviewed on their credentials, security and ethical oversight, with a stricter 'High-risk Use' tier for the most sensitive projects.
Why it mattersIt is an attempt to let real medical research use AI fully while still guarding against someone misusing it to design a bioweapon.
Anthropic ↗Quick hits
Smaller stories worth a glance.
- Google's SynthID Bio watermarks AI-designed proteinsDeepMind's new watermark hides a signature in AI-designed protein sequences and 3D structures without changing how the proteins work, aimed at helping screen dangerous DNA orders.Help Net Security ↗
- IBM lets companies run its AI coding tool on their own serversIBM launched a self-hosted version of Bob, its agentic software-development platform, so organisations can run it on-premises or in air-gapped systems instead of sending code to the cloud.IBM Newsroom ↗
- OpenAI says it disrupted a campaign to copy its AI's reasoningOpenAI disrupted a campaign that tried to extract its models' hidden 'reasoning' to help train a rival system, with a spike of 16,000 requests from more than 4,000 users on 24-25 July and related activity from over 15,000 users fully shut down by 28 July.OpenAI ↗
Deals & funding
Who raised money, and who bought whom.
Funding rounds 1
Investors put money into a company in exchange for a share of it.
Wellington backs ElevenLabs
- Raised
- $300M
ElevenLabs now valued at $22B
ElevenLabs sold $300 million in employee shares in a tender that values the voice AI company at $22 billion, double its February Series D valuation.
TRT. Rowe PriceEEQTGSGoldman SachsGGICOOTPPSVSapphire VenturesB&BDT & MSD
Andreessen Horowitz
LightspeedIICONIQDSD.E. ShawEEvanticDDISRUPTIVEAAlkeon+0 more
Stories are summarized in plain language; follow the link before quoting a figure. Story images are shown from the publishers' own sites.







