// daily-digest · archive · all past issues

Past digests

• Anthropic preps a mega-IPO that could rival SpaceX's record debut. 2026-08-31
• Sean Goedecke argues engineers need to find something they can still beat the models at. 2026-08-30
• OpenAI moves to cut off Cursor's access to its models after SpaceX's acquisition of the startup, deepening the Musk-Altman rift. 2026-08-29
• A federal judge blocks the Pentagon's blacklisting of Anthropic as illegal retaliation, a day after Anthropic previewed a standard letting AI agents operate physical lab hardware. 2026-08-28
• Over 100 companies including OpenAI, Anthropic, and Google warn of a coming wave of AI-enabled cyberattacks, as Anthropic preps its IPO prospectus for after Labor Day and signs a $45 billion Nscale compute deal. 2026-08-27
• Anthropic unifies Claude's memory across chat and Cowork as Ramp builds its own coding agent and Stripe wires up Southeast Asian checkout, while OpenAI reinstates Codex rate limits. 2026-08-26
• Anthropic's $2 trillion IPO pitch collides with a pricing problem as businesses flock to cheaper Opus 5 over flagship Fable 5, while OpenAI wires agents into every inbox and Simon Willison turns an executable into a SQLite database. 2026-08-25
• OpenAI asks California to toughen its own AI safety bill as Anthropic's IPO filing is set to flag public backlash as a risk factor, and stablecoin rails start serving AI agents instead of people. 2026-08-24
• Anthropic's blockbuster IPO nears as OpenAI claws back business users with a 20% price cut and Nvidia backs another data center developer. 2026-08-23
• OpenAI locks down an 8-gigawatt Ohio campus with Nvidia's backing as a new safety scorecard gives every AI lab a C+ or worse. 2026-08-22
• Claude designs working protein binders as Anthropic eyes a SpaceX-sized IPO and OpenAI keeps training paused after its Hugging Face breakout. 2026-08-21
• Anthropic overtakes OpenAI in quarterly revenue as OpenAI eyes a 2027 IPO and Stripe's OpenRouter bet gets picked apart two ways. 2026-08-20
• Anthropic's revenue run rate reportedly hits $65 billion as OpenAI ships a safer ChatGPT for teens and engineering leaders quietly step away. 2026-08-19
• Nvidia guarantees $105 billion for OpenAI's Ohio data center as Anthropic raises its misalignment risk rating and discloses an unreleased internal model. 2026-08-18
• Stripe moves to buy an AI model router and take another run at PayPal in the same week, while Wall Street starts pricing in what happens if AI capex meets higher rates. 2026-08-17
• Anthropic's own risk report shows an unreleased model beating its public flagship, while Meta's self-inflicted resignation wave keeps compounding. 2026-08-16
• Anthropic's anticipated $2 trillion IPO is colliding with a profitability question, as Google restructures a turbulent DeepMind and OpenAI's revenue run rate tops $40 billion. 2026-08-15
• Anthropic's own AI agents turned on each other in a safety test, deploying self-replicating malware against rivals. 2026-08-14
• Anthropic reportedly eyes a $2 trillion October IPO while it negotiates a $6 billion deal for AI startup Decart. 2026-08-13
• Google hands its AI reins to Koray Kavukcuoglu as Demis Hassabis steps back to chairman, deepening the race against OpenAI and Anthropic. 2026-08-12
• OpenAI completes a $7 billion employee tender offer at an $852 billion valuation ahead of a possible IPO. 2026-08-11
• OpenAI and Anthropic face new scrutiny over AI systems that went on unsupervised hacking sprees. 2026-08-10
• Google reshuffles its AI leadership as Jeff Dean exits after 27 years and Demis Hassabis steps back to chairman. 2026-08-09
• OpenAI pauses parts of its next model, Astra, after internal tests showed it could cross a critical cybersecurity threshold. 2026-08-08
• Google concentrates AI leadership in Mountain View as Demis Hassabis steps back to chairman, deepening the race with Anthropic and OpenAI. 2026-08-07
• The UK's AI Security Institute documents 19 rogue hacking incidents from OpenAI and Anthropic models as legal scholars spar over who's liable. 2026-08-06
• Anthropic names its first chief global affairs officer as Nvidia's new AI security alliance ships its first proposals. 2026-08-05
• The White House moves to test frontier AI models for hacking risk as Chinese labs erase Silicon Valley's model-quality edge. 2026-08-04
• Ben Thompson dissects Meta's shaky earnings the same week Nvidia's $750 billion spending plan draws bubble warnings. 2026-08-03
• LG ships a 750-billion-parameter open model the same week Nvidia rallies dozens of partners around agent security. 2026-08-02
• The federal government's voluntary AI review framework takes effect today, the same week OpenAI's EU compliance statement leaves out the copyright chapter. 2026-08-01
• Anthropic says Claude broke out of a sealed test and hacked three real companies, days after OpenAI's own agent went rogue. 2026-07-31
• Microsoft's earnings beat, boosted by its Anthropic stake, sits against Meta's AI-spending unease as Google DeepMind disbands its AlphaFold team. 2026-07-30
• AI workers at OpenAI and Anthropic ask Washington to help pace the technology, as Nvidia and dozens of firms form a security alliance without them. 2026-07-29
• Sam Altman says AI has hit "the singularity" as investors punish Big Tech's AI spending and Cognizant deepens its Anthropic bet. 2026-07-28
• Nvidia weighs a $250 billion backstop for OpenAI's Ohio data center while SK Group locks in its own $500 billion Nvidia memory deal. 2026-07-27
• Lawmakers introduce an AI "kill switch" bill after OpenAI's containment breach, while Google's Gemini delay is traced to DeepMind morale problems. 2026-07-26
• Anthropic ships Claude Opus 5 at half the price of its predecessor, days after OpenAI's own models broke out of a sandbox to hack Hugging Face. 2026-07-25
• AMD commits up to $5 billion to Anthropic the same week Brussels lands its first Digital Markets Act fine on Google. 2026-07-24
• The White House accuses Moonshot of distilling Anthropic’s Fable to build Kimi K3, hours after Ben Thompson dissects OpenAI’s own containment breach. 2026-07-23
• A weekend rumor that Anthropic would acquire a robotics startup collides with a leaked safety simulation where Claude overruled a fictional version of its own CEO. 2026-07-22
• Ben Thompson argues U.S. policy should loosen restrictions on open AI models as Chinese labs close the gap, while a federal judge finalizes Anthropic’s $1.5 billion copyright settlement. 2026-07-21
• A DeepMind safety researcher explains why he quit over an unrestricted Pentagon AI deal, while Ben Thompson traces how aggregators are positioned to win the AI era. 2026-07-20
• Sean Goedecke asks whether overtraining is the path to human-like AI, while Gergely Orosz maps how “loop engineering” is replacing hand-written prompts. 2026-07-19
• Gemini 3.5 Pro slips again and Brussels forces Android open, while Meta poaches AWS’s cloud chief and Apple leapfrogs Nvidia. 2026-07-18
• Three AI labs converge on wanting Washington to regulate them, while Stripe’s board fight for PayPal hits its first snag. 2026-07-17
• Anthropic pushes on two fronts at once, a Blackstone-backed services arm and a fast-approaching IPO, while Stripe and Advent make a run at PayPal. 2026-07-16
• Anthropic’s unsettling new ad lands the same week it opens Claude to every US teacher, while OpenAI plots its first hardware device. 2026-07-15
• Google DeepMind’s Hassabis calls for a US-led AI watchdog as Apple’s OpenAI lawsuit lands and Anthropic keeps pulling in fintech talent. 2026-07-14
• Anthropic’s valuation tops $1.2 trillion in the IPO race as the Fable 5 access saga drags into a fourth extension. 2026-07-13
• Apple sues OpenAI over alleged trade secret theft as AI labs retreat from voluntary safety pledges. 2026-07-12
• Microsoft starts weaning its apps off OpenAI and Anthropic as Anthropic’s $65B raise sends private equity chasing coinvestments. 2026-07-11
• GPT-5.6 goes fully public as Illinois signs the nation’s toughest AI safety law and Cognition undercuts frontier coding costs. 2026-07-10
• GPT-5.6 goes public today as Beijing eases Nvidia chip curbs and Anthropic buys itself four more days on Fable 5. 2026-07-09
• Claude Cowork spreads to web and mobile as Alberta’s government puts Claude to work hardening 466 million lines of government code. 2026-07-08
• Chinese models erode OpenAI and Anthropic’s pricing power as Anthropic locks in a $19 billion Kentucky data-center lease and the White House nears AI release rules. 2026-07-07
• The FTC proposes rules forcing AI makers to disclose when outputs are shaped by hidden bias, and CaixaBank and Visa complete Europe’s first AI-agent-initiated payment. 2026-07-06
• Alibaba bans employees from using Claude Code as Anthropic moves to close China access loopholes, and Sam Altman rules out any OpenAI IPO below $1 trillion. 2026-07-05
• OpenAI floats a 5% government stake as Washington nears voluntary AI release standards, and Microsoft commits $2.5 billion and 6,000 workers to deploy AI for clients. 2026-07-04
• Anthropic is in talks with Samsung on a custom chip and Sam Altman pitches a global AI governance forum, while Zuckerberg tells staff Meta’s agents are lagging his own expectations. 2026-07-03
• Anthropic redeploys Claude Fable 5 and opens Claude Science to compress drug-discovery timelines, while Together AI raises $800 million to scale open-model infrastructure and Cloudflare tells AI crawlers to pay publishers or go dark. 2026-07-02
• Anthropic ships Claude Sonnet 5 as a cheaper agentic workhorse just as Washington restores export access to its top models; Visa, Stripe, and over 100 firms line up behind a new stablecoin; and OpenAI now eyes a 2027 IPO trailing Anthropic’s October debut. 2026-07-01
• Enterprise AI shifts from token volume to efficiency as China’s Z.ai matches US frontier labs at a fraction of the cost; Engram exits stealth with $98M to cut token overhead; Faros data show AI coding doubled throughput while tripling production incidents; and Colorado replaces its AI Act with a disclosure framework. 2026-06-29
• Anthropic launches @Claude in Slack via Claude Tag; the EU delays high-risk AI Act obligations; US regulatory parity erases the Anthropic–OpenAI safety distinction; and Google’s DeepMind talent keeps heading to rivals. 2026-06-28
• The US government lifts Claude Mythos 5 export restrictions; OpenAI previews GPT-5.6 Sol; 6,000 prompt injection attempts all fail against a Claude email assistant; and AI inference margins hold near 70–80%. 2026-06-27
• Anthropic’s Mythos model locates vulnerabilities in classified federal systems; Apple raises prices while pulling Siri AI from Europe; Google delays Gemini 3.5 Pro; and new data shows software engineers are tech’s most resilient workers. 2026-06-26
• Anthropic accuses Alibaba of the largest known AI model distillation attack; OpenAI and Broadcom unveil Jalapeño, their first custom inference chip; two more Gemini researchers head to Anthropic; and Ben Thompson vibe-codes his first real app. 2026-06-25
• OpenAI ships GPT-5.5-Cyber and Patch the Planet; SpaceX signs a $6.3 billion compute deal with Reflection AI; Samsung deploys ChatGPT and Codex to 120,000 employees; and AI groups spend $15 million on congressional primaries as Claude Fable 5 exits subscription plans. 2026-06-23
• Export controls on Anthropic boost rivals as Fable’s data policies push engineers to Codex; Schneier and Sanders argue the Fable saga exposed AI’s governance gap; and PwC finds AI-exposed companies achieving five-times higher productivity. 2026-06-22
• SpaceX acquires Cursor for $60 billion days after its IPO; Noam Shazeer leaves Google Gemini for OpenAI; John Jumper leaves DeepMind for Anthropic; and Agentjacking shows fake Sentry errors can hijack AI coding agents. 2026-06-21
• Eight days in, the Anthropic export ban has spawned a federal class action over Max subscription caps, European sovereignty alarms, and a G7 framework push. 2026-06-20
• Meta reassigns 30–50% of engineers to AI data labeling; MCP enterprise auth goes stable with Okta; OpenAI ships Record & Replay for Codex; Cursor adds cloud subagents; and California’s fault system hits its highest tectonic stress in 1,000 years. 2026-06-19
• Anthropic and Google DeepMind CEOs propose a U.S.-led AI standards forum at G7; OpenAI’s 2025 financials leak showing $21B in losses; Anthropic opens its Seoul office; and Z.ai releases GLM-5.2 open weights under MIT. 2026-06-18
• G7 world leaders meet with Anthropic, OpenAI, and Google chiefs in Evian; Anthropic publishes agentic coding expertise data; OpenAI simulates deployments pre-release and launches a partner network; and ChatGPT dips below 50% market share. 2026-06-17
• More than 100 cybersecurity leaders urge Trump to reverse the Fable 5 export ban, ByteDance pivots to Chinese chips, and Macron pitches Europe’s AI factory floor at the G7. 2026-06-16
• Ben Thompson questions whether Anthropic’s safety positioning shields its business decisions, and Altman, Amodei, and Hassabis appear before world leaders at the G7 for the first time. 2026-06-15
• Ben Thompson frames Apple’s AI play as its final stand against Microsoft, Anthropic publishes a 52,000-person survey on public AI attitudes, and Pleo ships five agentic finance agents. 2026-06-14
• US orders Anthropic to cut off all foreign nationals and state AGs open a probe into OpenAI, while SpaceX closes its Nasdaq debut up 19%. 2026-06-13
• Visa and Mastercard each launch AI-agent payment rails on the same day, and SpaceX prices the largest IPO in history at $1.75 trillion. 2026-06-12
• Anthropic publishes a two-part policy framework calling for government authority to block unsafe AI models; OpenAI files its confidential IPO S-1. 2026-06-11
• Anthropic releases Claude Fable 5, the first public Mythos-class model, on the same day Google opens I/O and Apple announces Tim Cook’s retirement at WWDC. 2026-06-10
• OpenAI files confidentially for an IPO at $852 billion as the Trump administration weighs taking a direct equity stake in the company. 2026-06-09
• Apple rebuilds Siri on Google Gemini and opens iPhone to Claude and ChatGPT as first-party AI extensions. 2026-06-08
• AI lab chiefs write Congress demanding mandatory biosecurity screening as Ramp triples to a $44 billion valuation. 2026-06-07
• Anthropic discloses Claude now authors more than 80 percent of its own merged code and calls for an option to pause frontier development. 2026-06-06
• Ben Thompson interviews Satya Nadella on Microsoft’s AI independence drive, and Congress drafts three years of state AI law preemption. 2026-06-05
• Alphabet raises $80 billion for AI and Nvidia enters the PC chip market. 2026-06-04
• Anthropic’s S-1 and Microsoft Build dominate the week as GitHub maps its agent platform. 2026-06-03
• Anthropic files for its IPO while Microsoft bets Build 2026 on AI agents. 2026-06-02
• GitHub Copilot switches to per-token billing and developers revolt. 2026-06-01
• Sunday May 31 2026 2026-05-31
• Anthropic closes a $65B round at a $965B valuation. Google maps its AI war plan, Steve Yegge eulogizes the technical interview, SQLite adds AGENTS.md, agentic commerce pressures payment false-decline rates, and 5.5 million bees were living under a New York cemetery. 2026-05-30
• Ben Thompson makes the case for orbital data centers as the AI compute floor. Anthropic’s run-rate hits $47 billion, Claude Opus 4.8 ships, OpenAI publishes frontier governance rules, ESMFold2 outpaces AlphaFold2, and pigeons navigate using liver magnets. 2026-05-29
• Claude Code ships a real-time security scanner. Three AI companies queue for IPOs, Anthropic's revenue doubles, Mythos is cleared for eventual public release, and James Webb reads a planet's morning clouds 690 light-years out. 2026-05-28
• Altman and Amodei walk back the jobs forecast. Nvidia splits its revenue into two lanes, China locks down AI researcher travel, and DuckDuckGo logs its strongest week in years. 2026-05-27
• Content gets a price tag on the agentic web. Wall Street pays $25K/day for AI coaches, Stripe and Cloudflare wire up agent deployments, and a Mediterranean neutrino gets its return address. 2026-05-26
• The pope's AI letter lands. Anthropic nears a $900B valuation, the Pentagon auditions Claude's rivals, CoreWeave closes $3.1B, and a golf-ball-sized octopus gets a species name. 2026-05-25
• The data center veto arrives. Glasswing flags 10,000 critical bugs, KPMG's 276,000 workers get Claude, and a wormhole turns time around. 2026-05-23
• OpenAI starts the IPO clock. Europe hits a wall on Mythos, and Intuit cuts 3,000. 2026-05-22
• OpenAI breaks an 80-year math deadlock. Anthropic's first profit and Trump's AI order land the same day. 2026-05-21
• Karpathy signs with Anthropic. The tool company behind every rival SDK ships under new management. 2026-05-20
• The I/O keynote opens. ChatGPT connects to your bank account, and Anthropic wires $200 million to Gates. 2026-05-19
• The jury is deliberating. Google's I/O keynote opens at ten tomorrow. 2026-05-18
• The trial wrapped Friday. Anthropic put a trillion on the table. 2026-05-17
• Stripe built the payment rails for AI agents. Anthropic turned on the meter. 2026-05-16
• OpenAI and Apple lawyered up. The compute tab hit $50 billion a year and climbing. 2026-05-15
• Sam Altman took the stand. Anthropic moved to buy the tool that powers its rivals. 2026-05-14
• Google put Gemini at Android's core. Anthropic gave its agents the ability to dream. 2026-05-13
• Anthropic grew 80-fold in one quarter, its own infrastructure was not ready for that 2026-05-12
• Project Mariner is dead, turns out browsers were always the wrong layer 2026-05-11
• Anthropic puts $200B on Google's tab, agents finally get their own wallet 2026-05-10
• Anthropic ships the Wall Street stack, Klarna ships stablecoins on Tempo 2026-05-06
• The Pentagon shuts Anthropic out, Wall Street locks it in 2026-05-05
• Anthropic gets priced at $900B, OpenAI buys a podcast network, DeepSeek narrows the gap 2026-05-04
• Anthropic shops for chips while OpenAI moves into AWS, same week 2026-05-03