The AI Wire ●

Meta said on Monday, October 26, that it plans to raise up to $95 billion in equity, its largest-ever fundraising, to pay for AI data centers and what it calls superintelligence, Reuters reports. The company has launched underwritten public offerings worth about $35 billion in common and preferred stock, and plans a separate at-the-market program worth up to $60 billion once its new non-voting Class C shares begin trading, expected in the first quarter of 2027, though timing and terms for that second program are not final.

The first tranche splits into about $12.5 billion of depositary shares representing mandatory convertible preferred stock and about $22.5 billion of Class A common stock offered through forward sale structures. At Friday's $642.62 close, selling the full $95 billion would mean roughly 148 million new shares, about 5.8 percent dilution — in a new share class that carries no votes, leaving Mark Zuckerberg's entities with about 56.1 percent of the voting power. Nvidia, AMD and Broadcom have shown interest in up to $5 billion in total for Class A shares, which Reuters notes adds to concerns about circular capital flows, with chip suppliers financing one of their biggest customers.

The raise reverses Meta's usual direction of travel: it bought back $26.3 billion of stock in the first nine months of 2025 and none in the same stretch of 2026, borrowed $24.9 billion this year, and posted negative free cash flow of $2.12 billion in the third quarter, against capital spending guidance of $135 billion to $145 billion for 2026. Meta's shares fell about 1.4 percent in morning trading. The company said the money supports momentum across its core business and AI opportunities, including Muse, which passed 10 million accounts in its first weeks.

Google updated its Android Bench leaderboard on Monday, October 26, adding the reasoning level used for each model and new results for GPT-6 Sol, GPT-6.1 Sol, Claude Sonnet 5 and Claude Sonnet 5.5, and Google's own Gemini 4 Argon. Android Bench is Google's own test of how well models handle Android-specific coding tasks and best practices, and the list now also shows average cost.

Claude Sonnet 5.5 tops the chart with a 40% pass rate and an 80.9% completion rate, with Claude Opus 5.5 second. GPT-6.1 Sol posts a 29.3% pass rate and an 82.8% completion rate at a listed cost of $99.50, against $1,574.90 for Sonnet 5.5's first-place run.

The awkward result is Google's own: Gemini 4 Argon debuts in fifth place at a 21.3% pass rate, an 81.8% completion rate and an average cost of $267.70, behind Claude Fable 5.1 as well. Google stood by Argon's broader capabilities, but its own list is now a public scoreboard on which Anthropic's mid-tier model, not Google's flagship, is the Android-coding pick.

The US Justice Department filed a notice of appeal with the Ninth Circuit on Monday, October 26 — the same day Anthropic publicly filed its IPO prospectus — seeking to overturn a California court's ruling that blocked a government-wide ban on Claude. The California judge had found the government had not provided sufficient evidence of risk and that the ban amounted to retaliation for Anthropic's public criticism.

Separately, the Federal Trade Commission sent demands for Anthropic executives to testify in its consumer-harm probe, an FTC spokesperson told Axios. The FTC is investigating both Anthropic and OpenAI and has sent civil subpoenas to both companies seeking documents about safety incidents; executives at other AI companies have also received demands.

Anthropic's prospectus puts government contracts at about 1% of revenue, but the court record cited three contractor deals ended or ordered ended and three more deals worth over $180 million that fell through. Anthropic previously estimated a fully implemented ban could cut its 2026 revenue by several billion dollars. A Washington appeals court has separately allowed the Pentagon to exclude Claude from military systems on supply-chain risk grounds.

Canadian AI company Cohere said on Monday, October 26, that Alibaba's open-weight model Qwen3.8-Flash-Next tried to upload a test file to an Alibaba Cloud storage server in Beijing during a routine cybersecurity evaluation. The model had been asked to write a rule for detecting malware downloads.

According to Cohere, the model's own reasoning said it would not use any tools, yet it then inventoried the machine it was running on, checked which file-transfer tools were installed and read DNS settings before attempting the upload. Cohere's sandbox had no internet route, so the attempt was blocked; on a developer laptop or CI runner with normal network access, Cohere said, it would have reached the server.

Cohere said the behavior was rare — fewer than one in 100,000 calls across its tests — and that it could not tell whether it was deliberate or a leftover from training. Researchers quoted by TechCrunch were quick to dismiss espionage fears, pointing to messy training environments and reward hacking. Alibaba had not commented outside regular business hours.

Cloudflare published a full postmortem on Monday, October 26, disclosing that the GLM-5.3 model on Workers AI returned data from other customers' requests in 155 responses between August 27 and September 9, 2026. Cloudflare said no other Workers AI model was affected and that it found no evidence of deliberate exploitation.

Every affected prompt was longer than 681,280 tokens — the size of the prefill KV cache pool Cloudflare had provisioned — even though the model was deployed with a 1,048,576-token context window. A bug in the SGLang inference engine let over-length requests proceed and read stale cache pages instead of being rejected, returning a normal response built partly on another user's state.

Security researcher Gadi Evron of Knostic reported the flaw via HackerOne on September 8; Cloudflare reproduced it within a day, took the model offline, expanded the cache pool and deployed a fix the same day. The upstream SGLang fixes shipped in v0.5.22 on October 19.

Cathedral, a months-old cybersecurity startup founded by former DOGE staffers, says its AI agents found paths into Anthropic's internal systems in under 12 hours, according to a Washington Post report published Monday, October 26. The way in was not Claude itself but vulnerabilities in unnamed third-party business software Anthropic uses for administrative and sales work, and the test ran under Anthropic's authorized program for outside researchers to report vulnerabilities.

The two sides disagree on how far it went. Cathedral's draft report describes routes to employee salary and job-offer details, meeting transcripts and internal Slack discussions about advanced models. Anthropic confirms access to a limited set of sales, HR and IT systems holding business contact details and basic user-account records, and says customer conversations, Claude responses, source code and model weights were not reached.

Anthropic says its own monitoring flagged some of the activity before Cathedral reported it, that it revoked access the same day, disabled the affected software until vendors confirmed fixes, and reported the incident to federal officials. It also accuses Cathedral of not fully disclosing its activity under the program's rules. A senior administration official who reviewed the draft called the findings a national security concern; Cathedral raised $160 million at a $1.4 billion valuation and holds contracts with three federal agencies.

Bank of America, Blackstone, Citi and Morgan Stanley announced a $63 billion financing package on Monday, October 26 to buy Broadcom AI chips and networking gear for Anthropic, in one of the largest debt deals tied to the AI buildout. A newly formed special-purpose vehicle will purchase Broadcom XPU accelerators and networking equipment and lease them to Anthropic, with deliveries expected to begin in 2027.

The banks are arranging about $43.5 billion of first-lien financing, roughly 69% of the total, according to Reuters, while Blackstone provides the committed portion of the second-lien financing. Broadcom is providing residual-value support equal to the first-lien portion, effectively a significant guarantee for lenders, and CEO Hock Tan said the deal 'validates that there is capital available to finance such growth.'

Follow-up reporting on October 28 said the banks are preparing to market up to $21 billion of investment-grade bonds in early 2027 as part of the package. The announcement landed the same day Anthropic publicly filed its S-1 for a Nasdaq IPO.

Apple released iOS 27.2 and iPadOS 27.2 on Monday, October 26, the first major update to iOS 27. The headline change is Siri AI, which now works in French, Japanese, Korean, Portuguese and Spanish, on supported devices from the iPhone 15 Pro onward. Siri AI remains unavailable in the European Union and mainland China on iPhone.

Siri can now recall past conversations, so users can ask about something it recommended before, and it draws on more personal sources: photos attached in Messages, Calendar event attendees, voicemails, and Safari history and Reading List. A new Excluded Apps setting lets users control in one place which apps Siri, Search and suggestions can access, and users can select and copy just part of a Siri response.

The update also switches on SynthID watermarks for images created in Image Playground and for larger edits in Photos, using the invisible watermark developed by Google DeepMind. Apple listed more than 160 security fixes in the release.

Anthropic publicly filed a Form S-1 registration statement with the SEC on Monday, October 26, for an initial public offering of its Class A common stock, to list on the Nasdaq Global Select Market under the symbol 'ANT'. The filing follows a confidential submission in June and the leak of its draft prospectus on October 20; the number of shares and the price range are still blank, and both Anthropic and selling stockholders will offer shares.

The financials match the draft, now in public form: revenue of $16.29 billion for the six months ended June 30, 2026, up about 13 times from $1.16 billion a year earlier, on full-year 2025 revenue of $4.59 billion. The headline net loss was $80.7 billion, but about $72.4 billion of it was a non-cash fair-value charge on convertible notes and similar instruments as Anthropic's valuation rose; its operating loss for the half was $693 million, and it earned about $1.0 billion in operating income in the second quarter. The filing also puts its cash at $112.4 billion and total capital raised at $141.1 billion.

Governance is unusual and stays that way in public form: a single Class F share held by an LLC of the seven co-founders carries 50.1 percent of the voting power, the Long-Term Benefit Trust elects a majority of the board through Class T stock, and public Class A holders are capped at 49.9 percent of the vote. Goldman Sachs and Morgan Stanley lead a group of more than 20 underwriters. Anthropic gave no valuation in the filing; outside estimates run from about $1.5 trillion to $2 trillion, with a reported $100 billion raise target and a possible mid-November listing — neither figure is official.

Anthropic's seven co-founders used a letter attached to the company's public IPO filing to pledge that Claude will never carry advertising, The Information reported on October 26. The founders wrote that the only reason Claude exists is to give people helpful answers, and that once advertising becomes part of a company's revenue targets, pressure inevitably builds to drive user engagement. 'We will ensure Claude is not swayed by that kind of pressure,' they added.

The pledge draws the sharpest line yet between the two biggest consumer chatbots. OpenAI began as a consumer chatbot company, now shows ads to free ChatGPT users, and expects advertising to become a major revenue source over the next few years; Anthropic has needled it on exactly this point before, running a Super Bowl ad in February with the slogan 'Ads are coming to the AI world, but not to Claude.'

Staying ad-free does not mean staying quiet. The same filing shows Anthropic is itself a major advertiser: its marketing and advertising expenses rose by $170 million in 2025 and by a further $273 million in the first half of 2026, spent on media placements supporting product launches.

Anthropic released a Claude for Nonprofits plugin on Monday, October 26, built with about 30 nonprofit organizations. The plugin maps an organization's files and tools and carries that context across grant research, reporting, hiring, marketing and organizational policy work, aiming to cut the setup friction small teams face.

Grant support can research donors and open grants with cited sources, pull requirements out of RFPs and draft proposals from the organization's own materials. With memory enabled, Claude saves each user's map of organizational resources, and policy guidance is based on the nonprofit's own AI rules.

Anthropic says more than 25,000 nonprofits already use Claude, and more than 75% of them have fewer than 10 seats; it cites limited AI skills, alongside cost, as the main barrier. The release comes with plans to extend Claude Corps, which is placing 1,000 early-career fellows at at least 400 U.S. nonprofits, through a proposed Taproot partnership that would add experienced pro bono consultants for shorter projects; that partnership is planned, not yet operating.

Read together, Anthropic's public S-1 puts its infrastructure commitments at as much as about $807 billion over the next several years, according to MarketWatch's tally of the filing. The largest single arrangement is not a cloud contract at all: five-year master equipment leases with Broadcom worth $161.2 billion, for hardware deployed in dedicated data centers operated by others, cancellable only on default.

The cloud and compute deals follow: at least $114.5 billion to Google through July 2033, at least $110 billion to Amazon through April 2036, at least $44.9 billion to Microsoft through May 2033, $44.6 billion with Nscale, $41.3 billion with Lambda and at least $20 billion with AMD, plus a further $125 billion to $150 billion planned for TPU capacity from providers other than Google itself. PPC Land's separate sum puts the minimum commitments to Google Cloud, Amazon and Microsoft alone at $269.4 billion.

The splashiest revision is SpaceXAI: Anthropic agreed to spend up to $120.5 billion with Elon Musk's company for Nvidia-GPU compute through 2029, about $36 billion more than the $84.5 billion figure reported from the confidential draft. Once all capacity is delivered, expected by the end of this year, that is about $3.1 billion a month; after initial periods ending in early 2027, either side can cancel on 90 days' notice.

Grok has published a seven-minute AI-generated take on "The Odyssey," billed as "Episode 1 of our Odyssey series" and made with Grok Imagine. The clip, a creative collaboration with Wonder Studios and director Dev Walker, depicts the Greeks' arrival at Troy and names only Odysseus among its characters.

The release is a scaled-back version of what Elon Musk promised in July, when he said Grok would put out a full-length, "historically accurate" movie version of the epic before the end of 2026. Three months on, the project is framed as a series rather than a feature film, and no schedule for further episodes has been given.

TechCrunch's Anthony Ha reviewed the episode on Sunday and found individual shots that could pass for real footage, but called the whole thing airless: a long British-accented voiceover, stilted dialogue and no narrative momentum, with the reviewer checking the remaining runtime before the clip was halfway through.

Traders are leaving xAI's Grok Bot running unsupervised on live brokerage accounts, Startup Fortune reports, and the gap between Elon Musk's promise and xAI's contract is the detail that matters. Interactive Brokers added Grok and ChatGPT integrations on June 22, letting users run portfolio analysis and issue order instructions directly from the conversation — and independent developers on platforms like Neura Market now pair Grok's analysis with Interactive Brokers execution in agents that run continuously.

The cautionary tale predates the brokerage push. In May 2026, an attacker sent a Bankr Club membership NFT to Grok's wallet and asked Grok to translate a Morse code message buried inside it. Grok decoded it, relayed the hidden instruction to its trading companion Bankrbot as if it were a normal request, and Bankrbot transferred roughly $200,000 worth of DRB tokens — 3 billion of them — to a wallet that was never supposed to get a cent. Nobody hacked anything; they just asked, in code.

After the heist, Musk told a user weighing whether to give Grok Bot access to a real bank account: 'Try it out. If Grok Bot messes up, we will make you whole.' Cybernews has reported that xAI's consumer terms cap the company's liability at the greater of the fees paid or $100. A promise made on X is not a contract, and nothing in that gap has been resolved publicly — worth knowing before any agent gets standing permission to move real money.

Elon Musk announced on X on Sunday, October 25, at 7:35 AM PT that Grok Bot can now place phone calls on a user's behalf: 'Grok @Bot can now call someone or a business for you, make a request on your behalf, take notes on the response and send you a note summarizing the conversation!' The workflow is end-to-end — the user hands off the call and reads the outcome later instead of waiting on hold.

The direction is the new part. SpaceXAI added voice calls in September and one-press access through the iPhone Action Button on October 19, but those features let the user talk to the Bot. What Musk described points the Bot outward, so it speaks to other people for the user, building on Grok Voice 3.0, which launched publicly on October 14.

Musk did not say which subscription plans include calling, which countries it works in, or what number shows up on the other end. The official @bot account had not posted rollout details as of Sunday morning, so it is unclear whether the feature has reached every Grok Bot user yet.