The AI Wire ●

Israeli security firm Lava says it has identified 32 Nvidia Blackwell Ultra B300 accelerators — among Nvidia's most powerful data-center chips, and barred from sale to China under U.S. export controls — on monitoring systems linked to state-owned China Telecom, The Information reported on Thursday. Nobody raided a warehouse to find them: the chips surfaced through publicly reachable instances of Nvidia's DCGM Exporter monitoring tool, left on the internet without authentication.

The China Telecom systems were part of a much wider exposure. Lava's scanning found roughly 2,100 exposed DCGM Exporter hosts worldwide reporting more than 12,000 unique GPUs across nearly 300 organizations — H100s, H200s, B300s and consumer cards — with the exposed hardware valued at more than $100 million. Lava also disclosed a high-severity flaw in the same tooling, CVE-2026-47483, that could let an unauthenticated attacker crash GPU monitoring remotely.

The finding needs its caveat stated plainly: an IP address associated with a carrier is a network association, not proof of where hardware physically sits or who owns it — traffic can be proxied or tunneled — and Lava itself says the discovery raises questions about the chips' location and supply routes without proving an export-control violation. Even so, it is the first physical-evidence-style sighting in a diversion debate that has mostly run on paperwork, arriving alongside a separate October disclosure that a Chinese state-linked lessor financed more than 3 billion yuan of servers, including B300 systems, since August 2025.

Microsoft's AI diagnostic system, MAI-DxO, correctly diagnosed about 83% of complex medical cases in a peer-reviewed study published Thursday in Science — more than double the roughly 36% scored by generalist physicians working individually with reference materials. Microsoft AI CEO Mustafa Suleyman called the result a step toward what he calls 'medical superintelligence.'

The study, 'Sequential diagnosis with language models,' introduces a benchmark called SDBench that turns a case into a step-by-step encounter: the participant starts from a brief summary, then asks questions, orders tests and commits to a diagnosis. On 304 clinicopathological conference cases from the New England Journal of Medicine, MAI-DxO reached 83.2% accuracy (85.5% in peak configurations) against 36.1% for the physicians; on a second test of 325 emergency-department cases from Beth Israel Deaconess Medical Center, it scored 80.3% against 46.8%. MAI-DxO coordinates several AI agents that act like a panel of doctors, and it is model-agnostic — Microsoft tested it over models from OpenAI, Google, Anthropic, Meta and DeepSeek, with the best results on OpenAI's GPT-5 and GPT-5.5, and the orchestration layer alone lifted a GPT-5.5 baseline from 72% to about 87%.

The caveats matter: these are simulations over published case records, not a clinical trial, and cases this complex are normally handled by specialists or physician teams, who were not part of the comparison. Microsoft is now working with Beth Israel Deaconess to evaluate the system in a real hospital setting. 'The biggest challenge in healthcare delivery at the moment is the growing mismatch between demand and supply of medical expertise,' Dominic King, Microsoft AI's vice president of health, told TechCrunch.

Google and distributed-energy platform Voltus announced an expanded agreement on Thursday to build a 1-gigawatt virtual power plant in the PJM grid region, which serves 67 million people across 13 states and Washington, D.C. The deal adds 1,070 megawatts of flexible capacity to the 100-megawatt agreement the two companies signed in June, and the companies call it one of the largest distributed capacity portfolios deployed by a hyperscaler to date.

A virtual power plant aggregates capacity that already exists at homes and businesses — batteries, smart thermostats, cooling systems and industrial processes. When the grid is under strain, Voltus signals participating sites to reduce or shift their energy use, and participants receive bill savings and direct payments in return. Google funds the capacity as a way of bringing new supply to the grids its data centers draw from.

The announcement leans on a Brattle Group estimate that better grid utilization could save U.S. consumers more than $100 billion over the next decade. 'Expanding our initial commitment tenfold demonstrates our core effort to actively bring our own capacity to the grids that serve us,' said Ellen Zuckerman, Google's head of energy market development for the Americas. Voltus, which raised $225 million earlier this month and targets 20 GW of capacity by 2030, said the deal translates data center growth into direct economic benefits for homes and businesses.

Personal Intelligence in the Gemini app is expanding its Google Maps integration. Alongside the existing Search services connection, Gemini can now reference your Maps searches and saved places, and combine them with Calendar access to give tailored recommendations for your day.

Google's own example: ask which restaurant from a saved list is best after today's meetings, and Gemini checks where the last meeting is and finds saved locations near it. Personal Intelligence can be enabled from Gemini's personalization settings, and the expansion follows this week's rollout of the same feature into Android Auto.

Separately, following the launch of Gemini 4 Argon, Google is notifying AI Ultra subscribers that the new model will 'replace Deep Think soon,' suggesting they switch to Argon for deeper reasoning. The brief in-app notice does not make clear whether Deep Think — currently a thinking level on Gemini 3.1 Pro — is going away because that model is being retired, or only the mode is being phased out.

Google Cloud unveiled Agent Compute Engine on Thursday at the AGNTCon and MCPCon events. It is a compute service built specifically for running AI agents at scale, with each workload inside a hardware-isolated microVM that gets its own dedicated guest kernel and root access.

The isolation is the point: agents run code, call tools, and touch data on behalf of users, so Google is leaning on its Titanium IPU technology for added protection. The service is tuned for fast churn — agents can start quickly, pause, and resume later.

Its density trick is copy-on-write memory sharing: agents starting from the same base share memory until one changes something, which Google says has delivered benchmarks of 10 times the density on the same hardware. Google frames the service as part of its Gemini Enterprise Agent Platform, with security and governance for agents at the center.

Goldman Sachs announced on Thursday a $500 million commitment to expand its 10,000 Small Businesses program, headlined by a new Goldman Sachs Small Business AI Institute. The institute is partnering with Anthropic, Google, Meta and OpenAI to give entrepreneurs training and tools to integrate AI into their operations.

The institute will work on three fronts: delivering AI education to small-business owners across the country, publishing research on AI use by small businesses, and supporting AI-enabled expansions of access to capital. The wider commitment aims to deliver education, capital access and support services to another 10,000 businesses.

Nearly 20,000 entrepreneurs across all 50 states and four US territories have completed the program since it launched in 2010. 'Every major technology shift has rewarded the businesses that adapt most quickly,' CEO David Solomon said. 'This commitment makes sure the entrepreneurs who are creating America's jobs can lead in the AI economy.'

DeepSeek is wrapping up a funding round of about 100 billion yuan (roughly $15 billion), four people with knowledge of the matter told Reuters on Thursday — one of the largest funding rounds in China. Battery maker CATL and WeChat operator Tencent were among the biggest investors, with more than 15 investors in total, including new backers Midea, Meituan, Geely and GigaDevice.

Reuters reported the round values DeepSeek at 500 billion yuan before the new capital, double its original 50 billion yuan fundraising target. Bloomberg first reported the $15 billion figure. The round is expected to complete in the coming days, but one source cautioned that financial details and the investor lineup are not ironclad and could still change; DeepSeek, CATL and Tencent did not comment.

The raise lands two days after the US Department of Homeland Security ordered DeepSeek out of federal civilian agencies, and follows a June round of more than $7 billion at a $50 billion valuation. It is the rare Chinese-lab story that clears the Wire's bar: a raise of this size, aimed at AI infrastructure and open models, directly funds the strongest open-weight challenger to the US labs.

Anthropic's crackdown on distillation came with a price tag, Reuters reported on Thursday: the paying accounts the company has cut off accounted for between 5% and 7% of its revenue in the first nine months of 2026, according to investors briefed at an investor meeting in San Francisco last week. Anthropic alleges those accounts were not ordinary customers but Chinese AI labs quietly mining Claude's outputs to train competing models.

The technique, distillation, has a student model learn from a teacher model's outputs. Anthropic said in February that DeepSeek, Moonshot and MiniMax had generated more than 16 million interactions with Claude through roughly 24,000 fake accounts, in violation of its terms. Earlier this month it rolled out new defenses — access restrictions, proxy blocking and bans on tens of thousands of accounts — which it says have cut the reasoning data being siphoned by 90% to 99% (covered here on October 12).

Anthropic frames the practice as intellectual-property loss with national-security implications, and it is telling investors the crackdown has effectively stopped Chinese developers from distilling its models. The disclosure also hands the company an explanation for a month-to-month growth deceleration investors were already nervous about, weeks ahead of what could be the largest IPO in history.

Anthropic and formal-verification startup Theorem say Claude has done something that used to take experts years: it wrote mathematical proofs that real software will not crash. In research published on October 22, Claude proved that 239 test entry points across 123 open-source libraries from Google's OSS-Fuzz programme never crash on any input under 64 MB, working at the level of compiled RISC-V machine code.

The proofs run to roughly 180 million lines of Rocq, a proof language checked mechanically by computer, with a typical proof around 600,000 lines for a binary with a median of 18,000 instructions. Along the way Claude found bugs that trillions of fuzzing runs had missed, patched them, and then proved the patched code crash-free. Anthropic is publishing a subset of the proofs and the trusted components underneath them, so the checking does not depend on trusting the lab.

The bounds matter: the proofs cover single-threaded, static RISC-V binaries, not the x86 or ARM builds most software actually ships as, and 'formally verified' here means crash-freedom for the stated property, not bug-free in general. Even so, Anthropic frames it as the first large-scale use of AI for formal verification of production software at assembly level — arriving in the same week AI models helped surface the PHP2Shell flaws and Claude stripped unneeded code out of OpenSSH.

Federal lobbying disclosures for July through September 2026 show a step change in Washington: Nvidia spent $2.2 million, Anthropic $2.1 million and OpenAI $1.5 million on federal lobbying in the third quarter — a quarterly record for each company, The Verge reported on October 22. Anthropic's figure is roughly double its highest quarterly total from 2025, and OpenAI did not clear $1 million in a quarter until this year.

A broader count by the Project On Government Oversight, which adds outside lobbying firms' filings, puts the three companies' combined third-quarter spending at about $7.5 million, up 22% from the previous quarter and roughly 60% from a year earlier. The three have spent about $37 million on federal lobbying since the start of 2024, with roughly half of it coming in 2026 alone.

The filings cover the fights now shaping the industry: child safety legislation including the Kids Online Safety Act, chip export controls, cybersecurity evaluations, copyright, and who pays for data-center power — Anthropic's disclosure lists the Ratepayer Protection Act, which would push data-center energy costs onto developers. Lawmakers have introduced at least 99 AI-related bills in the current Congress. The sums are still modest next to the incumbents — Microsoft has spent at least $2 million a quarter since 2021 and Google regularly exceeds $3 million — but the trajectory is the story.

The Agentic AI Foundation (AAIF), the Linux Foundation group for open agent standards, announced four new hosted projects on Thursday at AGNTCon + MCPCon North America in San Jose: Agent Skills, contributed by Anthropic; Open Responses, contributed by OpenAI; OpenEnv, contributed by Hugging Face; and OpenSandbox, contributed by Alibaba. The additions take the foundation to 10 hosted projects in its first year, up from three at its December 2025 launch, joining MCP, AGENTS.md, goose, the Agent2Agent protocol, Agent Router and agentgateway.

Agent Skills is the headline transfer. Anthropic built the format while developing Claude Code, launched it in October 2025 and opened it as a cross-platform standard in December; a skill is a folder built around a SKILL.md file, with scripts and reference material an agent loads only when needed. Anthropic says more than 70 agent platforms, including ChatGPT, Cursor and GitHub Copilot, have adopted it, and the donation covers the standard's name and branding, its GitHub organization and the agentskills.io domain. Engineers from Anthropic, OpenAI and Google will maintain it in their personal capacities, ending Anthropic's sole leadership.

The four projects cover four layers of the agent stack: how agents gain skills, how applications talk to models across providers (Open Responses is OpenAI's open inference specification), how agents are trained and evaluated, and where agent code runs safely. 'A year ago, AAIF launched with three projects and a bet that agentic AI needed a neutral home to thrive. Today we host 10,' said Manik Surtani, the foundation's CTO and co-founder. The scale cuts both ways: the skills directory lists roughly 1 million published skills with nearly 280 million installs, and Snyk has already identified 76 malicious skills in one sample, which is why Anthropic has proposed version management, a formal proposal process and security guidelines for the standard.

Walmart announced on Wednesday that it is embracing agentic commerce, launching an Agent Gateway and related controls for customers who shop through AI agents. The move is the opposite of its biggest rival's: Amazon has recently begun blocking outside AI agents from making purchases on shoppers' behalf.

The Agent Gateway is Walmart's own MCP server — the Model Context Protocol was originally created by Anthropic — letting approved agents search Walmart and Sam's Club inventory and buy on a customer's behalf, without the customer having to hand the agent their personal information or purchase history. Walmart already offers integrations with Meta's Muse, ChatGPT, Gemini and other popular AI apps.

Agent Profiles, rolling out next month, give each agent a dedicated profile inside a customer's Walmart account, likened to adding an authorized user to a credit card. Customers can require approval for every purchase or set a spending limit, can delete a profile at any time, and do not need a Walmart+ membership. Walmart is also publishing terms for agents: they cannot exceed what the shopper intends, cannot access payment information, and non-compliant agents will be blocked — while Walmart's returns process still applies when an agent buys the wrong thing.

xAI's Grok Bot can now build reports, memos and dashboards, the official @bot account on X said on Wednesday. Users ask Bot for a page, then edit it together with Bot or have Bot keep the page up to date on its own, the post said, showing the feature in a 27-second video.

That is most of what is known. The announcement gave no availability, plan or pricing details, did not say which data sources dashboards can draw on or how often a self-updating page refreshes, and Grok Bot's public changelog had no matching entry when Tesla News Wire checked — the latest listed release was v0.71.2 from October 14. An October 11 release had already labeled canvases and documents as Page, with PDF download and sharing.

The feature follows SpaceX's third-quarter call on October 20, where president Gwynne Shotwell said Grok Bot had 800,000 paid weekly users and that business customers are its fastest-growing group — the audience living documents and dashboards are aimed at.

Two critical vulnerabilities in PHP — the language behind roughly 70% of websites, including WordPress sites — were publicly disclosed on Wednesday, with emergency patches issued by the PHP Foundation. The flaws, tracked as CVE-2026-109618 and CVE-2026-109619 and collectively nicknamed PHP2Shell, are rated 9.2 in severity, allow unauthenticated remote code execution, and have sat in PHP versions since version 5 in 2004.

AI found them first. Security firm Calif, working in OpenAI's new Patch the Planet initiative, used GPT-5.6 Sol to find one flaw in August, while Wiz independently found both using its Wiz Atlas agent together with Google's Gemini 4 Argon and GPT-5.6 Sol. OpenAI says it led a two-month coalition of maintainers, researchers and providers to prepare fixes, and is putting $1 billion in subsidized model access behind defender work of this kind.

The US National Security Agency warned that sites that cannot patch immediately should be pulled offline until they can. Patched releases are PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35; versions 8.1 and older are unsupported. Wiz notes the same AI capability that found the bugs also lowers the barrier to exploiting them, which is why the patch race is urgent.

Day 17 of Tibo Sottiaux's 28-day ship-or-reset run, which lasts until November 1, landed on Wednesday as a reset day. The post reads: bet you thought we forgot, but no, it is reset day today — a full Codex limit reset for users tracking the series.

The same post used the moment to remind developers to get Daybreak Blue access, OpenAI's vetted tier for security work with its frontier models, saying it only takes about 15 minutes to apply. Daybreak Blue is the access route behind the recent run of AI-assisted vulnerability work, including the PHP flaws disclosed this week.

At Thursday morning's sweep, Day 18 had not yet landed: the trackers still showed waiting for Tibo, with the challenge at 18 of 28 days.