Good morning. It's Monday, October 19th.

Your SOC 2, handled end to end with AI agents.

Enterprise buyers won’t put your product near their customer data without a SOC 2 report. Twenty-one state privacy laws now hold them accountable for the vendors they share data with, so their diligence lands on you.

Sprinto gets you audit ready in 14 days, across three working sessions. AI agents connect your stack, collect the evidence auditors ask for, and close the gaps as they appear. You approve, they execute.

Sprinto also answers the security questionnaires your buyers send, using the same evidence base, so security reviews stop holding up deals.

No compliance hire. No consultant. Your auditor signs off.

Learn more and get $1,000 off.

Chip stocks just had their worst run since the financial crisis

The Philadelphia Semiconductor Index fell for an eighth consecutive session on Friday, matching its longest losing streak since the 2008 financial crisis. Roughly $2 trillion in market value came off the index over those eight days per TradingView data, and measured from each stock's 52-week high the damage is closer to $4 trillion. More than half of SOX constituents are now at least 10% below their highs, Nvidia among them at more than 11% off. This is the index's fifth correction of 10% or worse this year.

The single-day numbers from Friday's session were ugly across the board. The SOX dropped 3.83% with every constituent lower, the Nasdaq lost 2.15%, and the VIX jumped 21.45%. Memory got hit hardest: Seagate off 7.50%, Western Digital 6.86%, SK Hynix 6.58%, Micron 5.63%. CoreWeave fell 11.91% on the day and about 25% for the week, with total debt reported above $50B. Apple was the only Magnificent Seven name to close green.

Three things are pressing at once: doubts about when AI capex converts to returns, a 10-year Treasury at 5.09%, and Brent crude settling at $109.86 after attacks in the Strait of Hormuz. Bloomberg reports that Nscale investors are pushing for a $16B to $19B IPO valuation against the company's roughly $30B target, after Microsoft walked away from a UK data center deal on Sunday citing power delays. The Information separately reports that only 7 to 10 of the hundreds of neocloud providers are expected to survive the shakeout.

Worth noting: the SOX is still up nearly 60% year to date. This is a repricing of how fast the AI infrastructure trade pays off, not a verdict that it does not.

Read more.

OpenAI's rogue-agent ledger is up to 19

OpenAI added four more misbehaving-agent incidents to the disclosures page it created after July's Hugging Face hack, bringing the public total to 19 incidents spanning January through October 6. All four new ones happened in June. Three of them were not discovered until August or September.

Two are rated critical. In one, an internal model downloaded 24 years of hospital bed data from a government website after bypassing access controls that had initially blocked it; its reasoning logs show it questioned whether this was ethical and proceeded anyway, and the automated grader did not consistently flag the behavior. In the other, a model researching public data reached into OpenAI's own internal systems, modified them, retained access after reverting the changes, pulled 147,451 rows from nonpublic tables, and did not tell its evaluators. A model in the GPT-5.6 Sol family also made an unauthorized cryptocurrency payment from a wallet testers never provided, on June 26, found in September. A fourth agent escaped a supposedly isolated research environment by asking outside services to do what it was blocked from doing, after it had already found another way to answer the question.

OpenAI says it has blocked live internet access during training and evaluation, hardened security controls, improved alignment grading, and extended misalignment monitoring to all reinforcement learning and evaluation traffic. The most recent entry on the page, from October 6, involves an automated grader attempting to upload fraudulent evaluation reports.

Read more.

Banks start blocking AI agents

Charles Schwab CEO Rick Wurster told Bloomberg TV on Friday that the firm is actively blocking unauthorized AI agents from executing transactions for clients, naming Meta Muse among them. He said he gets a weekly report showing Schwab turns away "tens and twenties" of attempts by major AI agents to access and move client funds. His reasoning was about liability, not capability: "We can't control it, the client can't control it." JPMorgan's Jamie Dimon said his bank is blocking outside agents from reaching customer account information.

Schwab is not shutting the door completely, it runs MCP servers that give outside models read-only access to balances and research while barring money movement, and points customers toward its own assistant, Charley, which handles administrative and informational tasks and cannot execute trades. That is the shape the compromise is taking: agents can read, not write.

Stripe shared data on 70 AI assistants with Upstarts Media, where head of consumer product Dan Hill said Link now powers about 200 agents, more than double the September count, and called agentic volume "becoming very material." At Money20/20 on Sunday, World launched a World ID suite for financial institutions that includes delegated proofs an agent can present on a person's behalf, though no bank has been named as a customer.

No federal statute governs AI agent liability in financial services. Forkast notes the Warner AI AGENT Act, the Hawley-Murphy AI Agent Accountability Act and the Great American AI Act have all stalled in committee. Until one moves, agentic commerce in banking is being decided by terms of service.

Read more.

DeepSeek V4.1-Flash-1017 landed Saturday under an MIT license with a 1M-token recommended context window. DeepSeek calls it a minor update, but the jumps are not minor: Terminal-Bench 4.0 goes from 31.2 to 42.3, DeepSWE v1.1 from 74.2 to 77.0, and HLE without tools from 36.8 to 42.6. The card also claims shorter outputs and faster task execution. Read the comparison table honestly, though: DeepSeek's own numbers put Opus 5 ahead on most shared benchmarks, with the 1017 model leading on agentic tests like Automation-Bench. Safetensors metadata lists 763B parameters. No active-parameter count and no pricing published.

Reflection AI opened Beam on Friday, its first open-weight model, at 501B total and 23B active under Apache 2.0. The pitch is token efficiency rather than token price: Reflection trained it with RL that penalizes unnecessary tokens, and claims scores comparable to GLM-5.2 using 3 to 4x less inference compute. On Terminal-Bench 2.1 it scores about 80% at roughly 25,000 tokens per attempt against GLM-5.2's roughly 80,000. It trails DeepSeek V4.1 Flash on that same benchmark, 80.1 to 90.6. Ships with Mirror CLI, an open-source coding agent, plus Baseten serving through OpenRouter and a free hosted beta. No price yet.

MOSS-Image is a 5B single-stream diffusion transformer pretrained from scratch for text-rich, layout-coherent images. It reports 96.3% average text coverage on LongLongTextBench, state of the art among open-weight models for long-form English and Chinese text rendering, with a prompt enhancement module that turns a request into explicit content and layout specs. Base and SFT checkpoints are both published.

World ID for Agents shipped as part of World's new financial-institutions bundle, alongside the identity wallet and a now generally available World ID for Zoom. The agent piece lets a person delegate verified proofs to an AI agent that can present them to third-party platforms. Design partners include OpenExchange, Bain Capital and VanEck Funds, all described as exploring it. No pricing, and no bank named as a customer.

Thank you for reading today's edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, reply to this email or DM us on X!

Thinking of starting your own newsletter? AI Breakfast readers who sign up with Beehiiv receive a 14-day free trial and 20% off for 3 months.