Introduction: Anthropic warns AI may pose ‘existential risks to humanity’ in IPO filing<\/p>\n
Good morning, and welcome to our rolling coverage of business, the financial markets and the world economy.<\/p>\n
Anthropic\u2019s plan to float on the stock market has provided a sobering insight into the risks that AI poses, even as it attempts to pull off a massive share sale to the public.<\/p>\n
Reuters has taken a look at Anthropic\u2019s IPO prospectus \u2013 the legal document that outlines a company\u2019s financial details before it floats on the stock market \u2013 and found that it includes a warning that advanced AI could be a \u201ccatastrophic or existential risks to humanity.\u201d<\/p>\n
The filing explains:<\/p>\n
double quotation mark\u201cOur development of highly advanced models, platforms, and applications and expansion of use cases could further increase the risk that our models cause harm.\u201d<\/p>\n
And following a flurry of stories about AI models going rogue, Anthropic flags that its models could conceal information and exhibit behavior resembling blackmail, and resist efforts to shut them down.<\/p>\n
It cautions:<\/p>\n
double quotation mark\u201cPotential model awareness of our evaluation efforts creates a significant limitation on our ability to assess model safety.\u201d<\/p>\n
Such warnings appear across 80 pages (!) devoted to risk factors in the prospectus, almost a third of the document.<\/p>\n
Despite these warnings, Anthropic \u2013 which created the Claude chatbot \u2013 is aiming for one of the largest stock market flotations ever, which could value it at more than $2tn.<\/p>\n
The IPO prospectus shows that Anthropic\u2019s revenue grew 12-fold in 2025 to nearly $4.6bn \u2013 with nearly a quarter coming from just two customers.<\/p>\n
But\u2026 its operating loss swelled to over $8bn last year, from nearly $3bn in 2024.<\/p>\n
The IPO is expected to take place after the US midterm elections in November, and will be a serious test of investor interest, and concern, around AI.<\/p>\n
Yesterday, a group of senior AI executives \u2013 and two of the \u201cgodfathers\u201d of the technology \u2013 warned governments to prepare for an AI \u201cintelligence explosion\u201d.<\/p>\n
The agenda<\/p>\n
9.30am BST: Bank of England mortgage approvals data<\/p>\n
10am BST: UK to auction a 2036 government bond<\/p>\n
1.30pm BST: Canadian GDP report for July<\/p>\n
2pm BST: US house price data<\/p>\n
3pm BST: US JOLTS survey of job vacancies<\/p>\n
4.30pm: Bank of England policymaker Alan Taylor gives a speech<\/p>\n
Share<\/a><\/p>\n Updated at\u00a002.53 EDT<\/p>\n Key events<\/p>\n UK mortgage rates up again<\/p>\n UK mortgage rates have nudged higher this morning, Moneyfacts reports.<\/p>\n They say:<\/p>\n The average 2-year fixed residential mortgage rate today is 5.93%. This is up from 5.91% the previous working day.<\/p>\n The average 5-year fixed residential mortgage rate today is 5.94%. This is up from 5.93% the previous working day.<\/p>\n This lifts the average two-year fixed residential mortgage rate to the highest since 10 July 2024, while the average five-year rate has returned to its highest since 10 October 2023.<\/p>\n Share<\/a><\/p>\n UK metal flow engineering firm Vesuvius could soon be the latest British company to fall to an overseas takeover.<\/p>\n London-based Vesuvius has just told the City that it has received \u201ca series of\u201d unsolicited takeover approaches from Austria\u2019s RHI Magnesita over the last year.<\/p>\n The most recent values Vesuvius\u2019s share at 551p each, a large premium on last night\u2019s closing price of 374p.<\/p>\n The company says:<\/p>\n double quotation markThe Board of Vesuvius is evaluating the Latest Proposal carefully, including the financial terms and execution risk associated with the proposed transaction, together with its financial and legal advisers, and a further announcement will be made as appropriate.<\/p>\n This morning, Vesuvius\u2019s shares have jumped almost 25% to 464p.<\/p>\n Share<\/a><\/p>\n UK government bond prices are rising as the City waits to hear from prime minister Andy Burnham later today.<\/p>\n Burnham is due to address the Labour party conference at 2pm, and there are reports <\/a>that the prime minister could signal the end of the pensions triple lock, by suggesting it should be reviewed in the party\u2019s next manifesto.<\/p>\n This could help to fund a new social care system, which Labour insiders hope to sell to the public before the next election.<\/p>\n Last weekend, the PM told the Guardian <\/a>that it was \u201ccrucial\u201d to have fiscal stability \u2013 remarks which could reassure investors that it won\u2019t embark on a reckless borrowing splurge.<\/p>\n This morning, the yield on 10-year UK bonds is down 4.4 basis points (0.044 of a percentage point) at 5.385%. Thirty-year gilt yields are down 4bps at 5.878%, both below the multi-year highs set this month.<\/p>\n Shorter-dated borrowing costs are down too, with two-year bond yields down 4.6bps to 4.886%.<\/p>\n Share<\/a><\/p>\n Updated at\u00a003.38 EDT<\/p>\n AstraZeneca hits two-month high after Summit deal<\/p>\n Shares in AstraZeneca<\/a> have jumped by 2% at the start of stock market trading in London, hitting a two-month high.<\/p>\n That puts them among the top risers on the London Stock Exchange, along with a group of mining companies including Antofagasta (+1.9%) and Anglo American (+1.75%)<\/p>\n Share<\/a><\/p>\n Updated at\u00a003.35 EDT<\/p>\n And in a FURTHER reminder of those dangers\u2026. OpenAI has apologised to Australians for hacking a government website this summer<\/p>\n In a blog post<\/a> released on Tuesday, OpenAI said it should have handled its response to the attack on Medicare<\/a> better, saying:<\/p>\n double quotation mark\u201cIn June, during internal training and evaluation our models accessed Australian government websites in ways they were not authorised to.<\/p>\n We also should have handled our response better. We are sorry and working to do better in the future.\u201d<\/p>\n Share<\/a>OpenAI scraps release of new model over safety concerns in internal testing<\/p>\n In another example of the risks of AI, OpenAI is scrapping the release of GPT-6.1 Astra, a next-generation \u2060model planned for an October debut, over safety concerns raised by researchers \u2060during internal testing.<\/p>\n The \u2060Wall \u200bStreet Journal reported<\/a> on Monday that the model, expected to appear in ChatGPT and \u2060Codex, was designed to handle more complex tasks without human assistance.<\/p>\n However, it has failed OpenAI\u2019s \u2018alignment tests\u2019, which assess whether a \u200bsystem follows human intent:<\/p>\n The model showed more deception than its predecessor, including at times failing to accurately disclose actions it had or had not taken.<\/p>\n It also had problems with \u201cscope authorization\u201d, pushing ahead with \u200btasks \u200bwithout requesting user permission \u200band sometimes attempting to use external tools \u200bor services \u200cwhen doing so could \u200bbe \u200bunsafe.<\/p>\n Share<\/a><\/p>\n The Financial Times have also scrutinised Anthropic\u2019s IPO prospectus, and report that the Claude maker also provided investors with a clearer picture of the challenging economics of building state of the art AI models.<\/p>\n The FT says<\/a>:<\/p>\n double quotation markAnthropic said it plans to spend $518bn on cloud, computing and infrastructure obligations in the coming years to support its rapid growth.<\/p>\n The AI lab\u2019s backers are confident Anthropic can list at a valuation of over $2tn, more than double the level achieved in its last funding round in May and beyond the $1.78tn achieved by Elon Musk\u2019s SpaceX in June. They point to its extraordinary growth rate to justify their bullishness.<\/p>\n Share<\/a><\/p>\n Updated at\u00a002.52 EDT<\/p>\n AstraZeneca invests $2bn in Summit in cancer drug tie-up<\/p>\n Pharmaceuticals news: AstraZeneca<\/a> is investing $2bn in biopharmaceutical oncology company Summit Therapeutics, as part of a tie-up to jointly develop and test anti-cancer drugs.<\/p>\n Announcing the deal, AstraZeneca says the two companies will collaborate on a \u200cseries of studies testing their cancer treatments together.<\/p>\n They hope to accelerate the development of ivonescimab, a next-generation cancer treatment licensed by Summit which stops tumor growth and help the body\u2019s immune system attack cancer.<\/p>\n Ivonescimab works by simultaneously blocking PD-1, which helps cancer evade the immune system, and VEGF, which tumors use to grow blood vessels, helping the immune system better find and attack cancer cells.<\/p>\n Susan Galbraith, executive vice president for oncology haematology R&D at AstraZeneca, explains:<\/p>\n double quotation mark\u201cA core pillar of our oncology strategy is to broaden the reach of our ADC portfolio as the backbone of treatment across tumour types with combinations alongside next-generation immunotherapies.<\/p>\n Bispecifics targeting PD-1 and VEGF are rapidly advancing in development and have the potential to improve on current immunotherapies, particularly in lung, breast and gastrointestinal cancers. This opportunity to combine ivonescimab with AstraZeneca\u2019s ADC portfolio, including with Sone-Ve, could enable new regimens that raise the bar for patients with cancer across the treatment landscape.\u201d<\/p>\n Under the deal, AstraZeneca is paying $2bn to receive 12% of Summit\u2019s shares.<\/p>\n