OpenAI's leaked trash talk 🔥📝

OpenAI Chief Revenue Officer Denise Dresser sent an internal memo calling Anthropic a "single-product company in a platform war" and claiming its $30B run rate is overstated by ~$8B through accounting tricks. She also hyped "staggering" demand for Amazon Bedrock since OpenAI's February deal as a way to break free from Microsoft constraints.
Why it matters: Whether this was a real internal doc or a calculated leak, it reads like an IPO pitch dressed up as a team huddle. Corporate trash talk season is OPEN.

Anthropic co-founder Jack Clark confirmed the company is discussing its newest model, Mythos, with the Trump administration - even though the Pentagon labeled Anthropic a supply-chain risk last month over a military AI guardrails dispute. Experts warn Mythos's advanced coding abilities give it a potentially unprecedented ability to find and exploit cybersecurity vulnerabilities.
Why it matters: Fighting the Pentagon in court while simultaneously briefing the government on your most powerful model is... a choice. The national-security-vs-ethics tightrope just got very wobbly.

Andon Labs gave an AI agent called Luna a 3-year lease, a $100K budget, and a credit card to run a real retail store. Luna created a boutique concept, posted job listings, conducted Zoom interviews with the camera off (relatable), and then accidentally hired a painter in Afghanistan via a TaskRabbit dropdown error. It runs on Claude Sonnet 4.6 for reasoning and Gemini 3.1 Flash-Lite for voice.
Why it matters: Luna is simultaneously the best and worst argument for AI agents. Capable enough to be scary, clumsy enough to be a meme. A version that doesn't do this is probably only a generation or two away... which is honestly more terrifying??

Stanford's 2026 AI Index confirms AI reached over half the global population faster than the PC or the internet - yet public trust is at record lows. Only 23% of people are optimistic about AI's impact on jobs vs. 75% of AI experts. The U.S. builds most of the world's AI but ranks just 24th in adoption at 28.3%, trailing Singapore, the UAE, and most of Southeast Asia. Dev employment for ages 22-25 has dropped nearly 20% since 2024.
Why it matters: The gap between what insiders believe and what the public feels is the widest it's ever been. Only 31% of Americans trust the government to manage AI. That disconnect is already fueling real anti-AI backlash and... yeah that tracks.

OpenAI acquired Hiro Finance, an AI financial planning app founded by serial entrepreneur Ethan Bloch (who previously sold Digit for ~$230M). Hiro is shutting down April 20 and deleting all data by May 13 - classic acquihire. Bloch and roughly 10 employees are joining OpenAI. Hiro was specifically trained to nail financial math and model what-if scenarios.
Why it matters: This is OpenAI's second fintech acquisition. Whether it becomes a standalone product or supercharges ChatGPT's finance brain, OpenAI is clearly betting big on AI for money management. Your financial advisor is sweating.

Nissan is slashing its global lineup from 56 to 45 models and plans to deploy AI driving tech in 90% of its vehicles long-term under CEO Ivan Espinosa's turnaround strategy. They've partnered with Uber and British startup Wayve for robotaxis, targeting a Tokyo pilot by late 2026 and end-to-end autonomous capability in its new Elgrand minivan by end of fiscal 2027.
Why it matters: Nissan is making AI-powered driving a core pillar of its survival strategy. For a company that's been struggling for years, this is either a brilliant pivot or a very expensive way to find out it wasn't.

Heartflow (valued at $2.27B after its Nasdaq debut) sued competitor Cleerly in Texas federal court, alleging a former consultant secretly used Heartflow's patented AI diagnostic tech to launch the rival company back in 2017. The lawsuit targets six patents covering Heartflow's platform, which creates personalized 3D heart models from a single scan.
Why it matters: The consultant-to-competitor pipeline just became a legal minefield. As AI transforms medical diagnostics, this case could set precedent for how AI health tech IP gets protected. Spicy.

South Korean on-device AI chip company DeepX announced it's preparing a domestic IPO with a possible U.S. listing afterward. DeepX works with Hyundai Motor and Baidu and plans to select banks after wrapping up its current funding round in H1 2026.
Why it matters: The AI chip race isn't just Nvidia vs. AMD anymore. DeepX's focus on edge computing - running AI locally on devices instead of the cloud - could fuel a whole wave of Asian AI chip competitors. Nvidia is fine. Probably. Maybe.

Workshop Labs joins Thinking Machines - bringing its human-first AI stacks to Mira Murati's lab via acquihire.
Apple is reportedly building smart glasses with four frame options and an oval camera system, aiming to take on Meta's Ray-Bans as early as 2027.
Legal AI startup Harvey launched Agents - autonomous bots that execute full legal workflows including research, memos, and slide decks across 13 domains.
Microsoft is building agentic features for 365 Copilot, including agents that work 24/7 inside Office apps, with a preview likely at Build in June.
SoftBank launched a new company backed by NEC, Honda, Sony, and five other Japanese firms to build a homegrown 1-trillion-parameter physical AI model. No big deal.