This week AI stopped being just a product story and became a governance story. Britain put public money behind a lab to defend against biological threats from AI, the two biggest names in cloud AI started selling each other's models, Anthropic disclosed a loss of more than $80 billion, researchers walked out of Google DeepMind over military work, and Anthropic promised 5% of itself to the public good. Here is what happened, with every claim verified against primary sources.
Key Takeaways
Britain set out around £600 million in AI security funding, including £100 million for a new AI Biosecurity Lab and £65 million for pathogen countermeasures, with OpenAI and Anthropic offering up to £200 million in model credits. (Reuters)
OpenAI models GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna are rolling out in private preview in Google Cloud's model garden, billed through Google Cloud Marketplace against existing cloud commitments. (Google Cloud Blog)
Anthropic recorded a net loss of more than $80 billion in the first half of 2026, versus $13 billion a year earlier, with $72 billion from accounting charges tied to convertible notes. (Reuters)
Google DeepMind researchers resigned over military AI work, with one saying staff fear retaliation, a claim Google denies. (Madrobot)
Anthropic will commit 5% of its shares to a new Public Benefit Fund, subject to stockholder approval, with 2% more possible for every 50% rise in value above a set baseline. (Anthropic)
1. Britain Funds an AI Biosecurity Lab: £100M and a Bigger Security Package
Britain set out around £600 million (about $800 million) of funding largely geared toward defending against health and security threats from AI, reports Reuters. The biggest chunk goes to a new biosecurity initiative: £100 million for a laboratory using AI to detect biological threats, plus £65 million to develop countermeasures against pathogens.
AI minister Kanishka Narayan said the aim is for the AI Biosecurity Lab to become a major global institution bringing together world-leading scientific expertise to protect people from emerging biological threats. The lab is described as the world's first organisation dedicated solely to AI biosecurity, building on Britain's broader Biological Security Strategy.
The private sector is contributing in kind rather than cash. OpenAI and Anthropic will offer up to £200 million of model credits to the biosecurity initiative, while Google DeepMind and London-based AI life sciences startup Basecamp Research will offer other assistance. The wider package also includes an extra £120 million for Britain's existing AI Security Institute and £130 million for cyber defence projects for critical infrastructure, including a National Cyber Shield and a National Exposure and Real-Time Vulnerability Centre.
The signal is unmistakable: governments now treat AI biosecurity as infrastructure, not as a research topic. A lab with a public mission, frontier model credits, and a life sciences partner is a bet that detection and countermeasures must keep pace with the models themselves.
2. OpenAI Models Arrive on Google Cloud: Rivals Selling Each Other's AI
In one of the week's stranger headlines, Google Cloud announced it is bringing OpenAI models to its customers, reports the Google Cloud Blog. GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna are rolling out now in private preview to select enterprise customers in Google Cloud's model garden.
The commercial terms explain why this makes sense for both sides. OpenAI models are transacted through Google Cloud Marketplace, appear on a single Google Cloud invoice, draw down existing cloud commitments, and count toward Flexible Savings Plans. Coming soon, according to Google: OpenAI models in Gemini for business and in Google Antigravity, its coding tool.
For buyers, the practical effect is choice without a second procurement process. A company already committed to Google Cloud spend can now use OpenAI models against that commitment. For the industry, it is a reminder that the cloud wars and the model wars are not the same war, and the winners in one may happily rent from the winners in the other.
3. Anthropic's $80B Loss: Accounting Charges, Not a Cash Burn Story
Anthropic recorded a net loss of more than $80 billion in the first half of 2026, reports Reuters, citing a Wall Street Journal report based on a draft IPO prospectus. That compares with a net loss of $13 billion in the same period a year earlier.
The headline number needs its footnote. About $72 billion of the loss resulted from accounting charges tied to changes in the value of convertible notes and other instruments issued to raise funding, not from day-to-day operating spending. Reuters also notes that Anthropic alone racked up more than $12 billion in operating expenses last year, which is the more useful gauge of its real cost base.
Read together with the next story, the picture is a company preparing for public markets while trying to lock in its public-benefit identity before investors arrive. Big accounting losses from convertible notes are common on the road to an IPO. What matters for readers is not the $80 billion figure alone, but what it is made of.
4. DeepMind Researchers Quit Over Military AI: Safeguards in Question
Two Google DeepMind researchers resigned over military AI work, in a BBC investigation reported on October 21, reports Madrobot, citing BBC interviews. London-based research engineer Sofia Liguori resigned after three years, saying that realising her company is pushing more and more for AI systems in military operations made her feel ashamed to work at Google. AI researcher Andreas Kirsch left earlier in October, saying Google's governance and oversight are not sufficient or meaningful enough to deploy AI safely and responsibly in military and classified settings, and that Google seems to have no right to control or veto how the Pentagon uses its AI.
Liguori also said staff are scared of retaliation. Google told the BBC it has strict workplace policies in place to prevent retaliation of any kind.
The resignations land in the same week as Britain's biosecurity funding, and the contrast is instructive. One story is a government funding AI for defence against biological threats. The other is researchers questioning whether a private lab's safeguards hold when its technology enters classified military use. Both ask the same question from opposite sides: who controls what AI is used for once it leaves the lab.
5. Anthropic Commits 5% of Itself to Public Benefit: A Fund, Not a Charity
Anthropic announced it will commit 5% of its shares to a new Public Benefit Fund, subject to stockholder approval, reports Anthropic. The board may add 2% more for every 50% increase in the company's value above a set baseline, at its discretion.
The structure is deliberate. The fund is established as an LLC rather than a 501(c)(3) charity, to give adjustability between different forms of giving. None of the Anthropic equity held by the LLC will be used for lobbying or political giving, in the US or abroad. Combined with co-founder and employee pledges, Anthropic says roughly 20% of its shares will be dedicated to the public good.
This is the other half of the IPO story. A company heading toward public markets, with an $80 billion accounting loss in its prospectus, is also writing a public-benefit commitment into its capital structure. Whether 5% proves meaningful depends on Anthropic's future valuation, but the mechanism, more shares as value rises, is designed to scale with success rather than stay fixed.
How the Week's Stories Stack Up
Aspect | Story | Key Number | Source |
|---|---|---|---|
Public funding | UK AI Biosecurity Lab | £100M lab, £65M countermeasures | Reuters |
Cloud rivalry | OpenAI on Google Cloud | 3 models in private preview | Google Cloud |
Company finances | Anthropic H1 2026 loss | Over $80B, $72B accounting charges | Reuters / WSJ |
Workplace dissent | DeepMind resignations | 2 researchers named | BBC via Madrobot |
Corporate structure | Anthropic Public Benefit Fund | 5% of shares, plus 2% steps | Anthropic |
The thread across all five: the AI industry is negotiating its social contract in public. Governments are funding defences, rivals are sharing storefronts, a leading lab is disclosing IPO-era losses while pledging shares to the public, and its own researchers are testing whether internal safeguards mean anything in classified settings. Last week was about the machinery around the model. This week is about who that machinery answers to.
If you like quick video takes on how new open AI projects work, browse the repo signals for 30-second overviews.
Frequently Asked Questions
What is the UK AI Biosecurity Lab?
The AI Biosecurity Lab is a new British government laboratory funded with £100 million to use AI to detect biological threats, plus £65 million for countermeasures against pathogens. OpenAI and Anthropic will offer up to £200 million in model credits, and Google DeepMind and Basecamp Research will offer other assistance.
Can I use OpenAI models on Google Cloud?
Yes, in private preview for select enterprise customers. GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna are rolling out in Google Cloud's model garden, billed through Google Cloud Marketplace against existing cloud commitments. Availability in Gemini for business and Antigravity is coming soon.
Did Anthropic really lose $80 billion in cash?
No. Anthropic recorded a net loss of more than $80 billion in the first half of 2026, but about $72 billion came from accounting charges tied to convertible notes and other funding instruments. Its operating expenses last year were more than $12 billion, a separate and more cash-relevant figure.
Why did DeepMind researchers resign?
Sofia Liguori and Andreas Kirsch resigned over Google's military AI work, saying governance and oversight are insufficient for military and classified settings. Liguori said staff fear retaliation; Google told the BBC it has strict policies against retaliation.
What is Anthropic's Public Benefit Fund?
It is a new LLC that will hold 5% of Anthropic's shares, subject to stockholder approval, with 2% more possible for every 50% rise in value above a baseline. Its equity will not be used for lobbying or political giving, and combined with other pledges, roughly 20% of Anthropic shares will be dedicated to the public good.
Conclusion
This week reframed AI news as accountability news. Britain funded a lab to catch biological threats early, Google and OpenAI proved rivals can share a storefront, Anthropic showed both its IPO accounting and its public-benefit pledge, and DeepMind researchers put their names to a warning about military safeguards. The models will keep getting faster and cheaper. The harder question, who they serve and who restrains them, is now being answered in funding packages, prospectuses, and resignation letters.
Read this article in Bahasa Indonesia: Berita AI: 24 Oktober 2026.