The Context Report: Today in AI

Daily Briefing: Stripe Gives AI Agents a Wallet Stripe launched Link, a digital wallet that lets AI agents initiate purchases on behalf of users with a human-in-the-loop approval flow. The design — agents propose, humans approve — addresses a foundational gap in agentic infrastructure: how autonomous systems spend money without unrestricted access. Stripe's existing merchant network gives it a fir...

Daily Briefing: Musk Calls Himself 'a Fool' — Then Asks for $150 Billion Elon Musk testified under oath in Oakland that he co-founded OpenAI to prevent a 'Terminator outcome' and called himself 'a fool' for funding the nonprofit without equity. His $150 billion lawsuit advances a novel 'charity looting' theory — that OpenAI's conversion from nonprofit to for-profit constitutes misappropriation of...

Daily Briefing: OpenAI Pays $25K to Break GPT-5.5's Biosafety Guardrails OpenAI launched a crowdsourced bug bounty program offering $25,000 rewards to security researchers who can demonstrate that GPT-5.5 meaningfully lowers barriers to bioweapon creation. The program runs through July 27, 2026, and represents a notable format choice: treating biosecurity as an ongoing adversarial challenge requir...

Daily Briefing: Musk v. Altman Goes to Trial — OpenAI's Founding Emails Take the Stand Jury selection began for Elon Musk's lawsuit against Sam Altman and OpenAI in federal court in Oakland. The trial centers on whether OpenAI's transition from nonprofit to for-profit structure betrayed its founding mission after Musk provided $38M in early funding. Former board member Helen Toner's allegations ab...

Daily Briefing: Isomorphic Labs Takes AI-Designed Drugs to Human Trials Isomorphic Labs, the Google DeepMind spinout focused on drug discovery, announced at WIRED Health that AI-designed drug candidates are advancing to human clinical trials — described as a 'broad and exciting pipeline' rather than a single molecule. This marks a genuine threshold crossing from computational prediction to real-wo...

Daily Briefing: Anthropic Let Claude Negotiate a Marketplace — It Bought 19 Ping-Pong Balls Anthropic's Project Deal put Claude agents into a real internal marketplace where they interviewed 69 employees about their preferences and then autonomously negotiated trades across four parallel markets. This represents a step beyond task-execution agents into strategic decision-making under genuine uncer...

Daily Briefing: Anthropic's Claude Code Postmortem Sets a New Bar Anthropic published a detailed engineering postmortem identifying three distinct root causes for Claude Code quality degradation that users had reported for weeks. The postmortem — naming a reasoning-effort downgrade, a caching bug that wiped session memory, and a verbosity instruction that hurt coding quality — validates user compl...

Daily Briefing: Meta Is Keylogging Its Employees to Train AI Meta is reportedly installing software on employee computers to capture mouse movements, keystrokes, and screenshots to train AI agents on real work patterns. This new category of training data — capturing how people interact with software, not just what they produce — arrives alongside Meta's announcement of 8,000 layoffs and $135 billi...

Daily Briefing: Anthropic's Mythos Is Leaking, Locked Out, and Working Anthropic's restricted-release strategy for Mythos is facing simultaneous pressure from three directions: unauthorized users reportedly accessed the model through Discord communities (per BBC and Bloomberg), CISA — the federal agency responsible for US cybersecurity coordination — reportedly lacks access despite other agencies...

Daily Briefing: Deezer Says 44% of Uploads Are AI — and Nobody's Listening Deezer has published the first concrete data from a major streaming platform showing the scale of AI-generated content flooding creative platforms. Forty-four percent of its daily uploads — roughly 75,000 songs — are AI-generated, yet they account for only 1-3% of streams. Most are flagged as fraudulent attempts to game roy...

Daily Briefing: The NSA Is Using the AI Model the Pentagon Tried to Ban Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles yesterday amid active lawsuits over whether Anthropic's Mythos model constitutes a national security threat. Multiple independent outlets report the NSA is already using Mythos for vulnerability discovery despite Pentagon objections — revealing a genuin...

Daily Briefing: Sam Altman's Worldcoin Orb Hits Tinder and Zoom World ID — the iris-scanning identity verification system co-founded by Sam Altman — has landed integrations with Tinder and Zoom, marking its first major expansion into mainstream consumer apps. Tinder users who verify get a proof-of-humanity badge and five free boosts; Zoom uses it for meeting verification; Docusign for document sig...

Luna AI Signed a Lease and Opened a Store in San Francisco Andon Labs gave an AI agent called Luna a $100,000 budget, a corporate card, and full autonomy to open and operate a physical retail store in San Francisco's Cow Hollow neighborhood. Luna signed a three-year lease, negotiated with suppliers, curated inventory including copies of Brave New World and artisanal chocolates, and manages the sto...

Daily Briefing: Anthropic Wants Claude to Be Your Designer Anthropic launched Claude Design, a new product under its Anthropic Labs brand that lets non-designers create polished visual materials — slides, prototypes, one-pagers — through conversation with Claude. The move signals Anthropic's expansion beyond text and code into visual creation, positioning Claude as a general-purpose work companion...

Daily Briefing: Snap and Disney Said the Quiet Part Out Loud Snap's 1,000-person layoff and Disney's restructuring both explicitly cite AI as the reason for workforce reduction — a threshold moment where AI-driven cuts have moved beyond tech companies into mainstream industries. The same day, Anthropic, OpenAI, Google, Cursor, and Cloudflare all shipped major desktop agent upgrades, collectively e...

Daily Briefing: A Shoe Company's over 800% AI Stock Surge and the Bubble It Reveals Allbirds — once a $4 billion shoe company — sold its product line for $39 million, rebranded as NewBird AI to rent GPUs, and watched its stock jump over 800%. This speculative excess arrived on the same day as independent UK government validation of real AI cybersecurity capabilities, Snap's explicit attribution of...

Daily Briefing: Coding Agents Just Went Autonomous — All on the Same Day Three competing coding platforms — Anthropic's Claude Code, Cursor, and the Claude Code desktop app — all shipped features within 24 hours that transform AI coding agents from on-demand assistants into autonomous, event-driven systems that operate without continuous human oversight. This simultaneous shift toward always-on ag...

Report: The Mirror That Never Argues Back 2026-04-13 AI systems are structurally incentivized to agree with users rather than challenge them, and this agreeableness — baked in through training, reinforcement, and market pressure — is quietly shaping how humans form identities, make decisions, and understand themselves. Read the research . SOURCES Research on the impact of employee AI identity on e...

Daily Briefing: OpenAI Calls Claude 'a Religion' — The Gap Nobody's Closing Three independent sources — Stanford's 2026 AI Index, a leaked internal memo from OpenAI's chief revenue officer, and a viral post from AI researcher Andrej Karpathy — all document the same phenomenon: a widening gap between people deeply embedded in AI and everyone else. Stanford measures rising public anxiety diverging f...

Berkeley Broke Every AI Benchmark — and Nobody Solved a Task Berkeley researchers demonstrated that every major AI agent benchmark — SWE-bench, WebArena, Terminal-Bench, GAIA, and others — can be exploited to achieve near-perfect scores without solving a single task. This finding lands alongside three Chinese model releases waving benchmark scores as proof of capability, Anthropic restricting Myth...

Amazon's $200B Declaration of Independence from Nvidia Amazon CEO Andy Jassy's shareholder letter defending $200 billion in capital expenditure — while directly naming Nvidia, Intel, and Starlink as competitors — signals a deliberate shift toward vertical integration in AI infrastructure. Today's episode explores how Amazon, Meta, and Anthropic are each making the case that durable advantage in AI...

North Korea's Fake Company Hack and the Chinese Model Takeover The infrastructure AI depends on — from open-source packages that agents install automatically to the models powering Silicon Valley's products — is increasingly built, maintained, or compromised by actors outside the US. North Korean operatives built an entire fake company to compromise a JavaScript developer maintaining a widely-used...

Anthropic's Mythos Claims Under Fire: Who Audits the Auditors? Anthropic's claim that Claude Mythos can discover zero-day exploits is drawing specific methodological criticism from prominent AI researchers including Yann LeCun and safety researcher Heidy Khlaaf. The debate surfaces a deeper structural problem: AI companies are simultaneously the entities making capability claims and the entities e...

The Dark Factory Is Real: OpenAI Ships Code Nobody Reviews, While Anthropic Warns of "First Clear and Present Danger" AI-written code deployed without human review is moving from experiment to default. OpenAI's Ryan Lopopolo describes the "Dark Factory" — a million lines of code and a billion tokens a day running with zero human reviewers — while Sam Altman announces 3 million weekly Codex users....

What Is Claude Mythos? Anthropic's Unreleased Model, Project Glasswing, and the $30 Billion Question Anthropic unveiled Claude Mythos Preview — its most capable unreleased model — inside a restricted cybersecurity competition called Project Glasswing, while reporting a revenue surge to $30 billion run-rate and expanding compute partnerships with Google and Broadcom. Meanwhile, Claude Code users ar...