An AI just cracked an 80-year-old geometry problem, Anthropic is weeks from its first profit, and the White House is quietly rethinking oversight it once dismissed.
OpenAI's AI solved an 80-year-old geometry problem. Autonomously.
One of OpenAI's general-purpose reasoning models independently cracked a famous geometry problem that had resisted mathematicians since the 1940s, without being specifically trained to do so. The AI reasoned through it the way a mathematician would, step by step, arriving at a proof that checks out.
Why it matters: This is not a benchmark score. It is original mathematical discovery. If AI can do this in geometry, the path to real breakthroughs in medicine, materials science, and engineering becomes much shorter. This is the kind of milestone researchers have been waiting for.
Anthropic is heading for its first profitable quarter, two years early
Anthropic is on track to post its first ever profitable quarter, with Q2 2026 revenue set to top $10.9 billion: more than double the previous quarter. The $559 million estimated operating profit arrives two years ahead of internal projections. The company also just signed a $1.25 billion per month compute deal with SpaceX's Colossus supercomputing infrastructure to power its next models.
Why it matters: The story of AI being expensive and unprofitable is changing fast. When the company building some of the most safety-focused AI in the world starts printing money, it shifts what is possible, and who gets to build what next.
The White House is quietly rethinking its hands-off stance on AI
The Trump administration, which had resisted AI regulation, is now reportedly considering oversight requirements for frontier AI models. The shift was triggered by concerns about Anthropic's Mythos model (which can identify and exploit cybersecurity vulnerabilities) raising national security questions that are hard to ignore. The White House is exploring a framework that would give government early access to powerful AI before public release.
Why it matters: When even the administration most opposed to tech regulation starts asking for AI guardrails, the direction of travel becomes clear. Expect some form of federal AI oversight in the US before the end of 2026.
EU AI Act transparency rules arrive in August. The countdown is on.
Starting August 2, 2026, the EU AI Act's Article 50 requires that AI-generated content (text, images, audio, video) be marked in a machine-readable format and disclosed to users at first interaction. Platforms operating in Europe must comply or face penalties. The rule covers any AI-generated content served to EU users, regardless of where the company is based.
Why it matters: This is the first major binding rule that requires AI content to be labelled at scale. It sets a template that India and other countries are watching closely as they draft their own AI governance frameworks.
India: 58,000 GPUs live, $126B AI opportunity, and skilling is the gap
India's IndiaAI Mission has provisioned more than 58,000 GPUs, with the latest tranche of 20,000 now confirmed. The country's AI opportunity is valued at $126 billion by 2030, driven by enterprise AI growing from $11B to $71B and consumer AI from $13B to $55B. At the same time, India's AI skill penetration is already 2.5x the global average, with government skilling programs running under the SOAR framework to prepare students and workers.
Why it matters: India has the infrastructure and the numbers. The window to build skills, products, and careers in this ecosystem is wide open right now, not in five years.
Google I/O ripples: Gemini Spark, Daily Brief, and SynthID are rolling out now
Following last week's Google I/O, several announced features are now live or in rollout. Gemini Spark (a personal AI agent that takes actions in Gmail, Calendar, and Tasks on your behalf) is starting to reach users. Google's Daily Brief feature builds a personalised start-of-day digest from your inbox and schedule. SynthID, Google's tool for detecting AI-generated content, has now been used 50 million times and is expanding to Search and Chrome.
Why it matters: These are not demos. They are in people's hands. The shift from AI as a chatbot to AI as an operating layer for your day is happening this month.
OpenAI is targeting a September 2026 IPO
OpenAI is preparing a confidential SEC filing and is targeting a public debut as soon as September 2026, making it potentially the largest tech IPO in years. Anthropic is also watching the IPO window. When the companies building the most powerful AI in the world go public, the accountability rules change: shareholders, regulators, and the public get a new lever. The pricing of that first day will tell you a lot about how the world values AI right now.
Deepfake job fraud + California AI laws + the DEFIANCE Act
Experian's 2026 Fraud Forecast names deepfake job candidates (synthetic applicants passing real-time video interviews) as a top emerging threat to hiring. Separately, a wave of California AI laws take effect in 2026 covering generative AI developers, healthcare AI tools, and algorithmic pricing. And the US DEFIANCE Act, which lets victims of non-consensual sexual deepfakes sue creators for up to $250,000, is moving toward final passage.