The Morning Brief — April 6, 2026
OpenAI Raises $122 Billion to “Accelerate the Next Phase of AI”
$122 billion. With a B. OpenAI has closed what might be the largest private funding round in history, earmarked for frontier AI research, next-gen compute, and meeting exploding demand for ChatGPT, Codex, and enterprise products. At this point, the company isn’t just building AI — it’s becoming a sovereign wealth fund that also ships chatbots.
Anthropic Sure Has a Mess on Its Hands Thanks to That Claude Code Source Leak
Anthropic accidentally released Claude Code’s source code, and now — because the internet is the internet — hackers are reposting it bundled with bonus malware. The timing is exquisite, given that Anthropic is currently the hottest trade in private markets and reportedly eyeing an IPO. Nothing says “we’re ready to go public” like your proprietary code doing laps on sketchy forums with a malware chaser.
Meta Pauses Work With Mercor After Data Breach Puts AI Industry Secrets at Risk
Mercor, a major data vendor used by multiple AI labs, suffered a breach that may have exposed sensitive details about how those labs train their models — the kind of information that represents years of competitive advantage and billions in R&D. Meta has paused the relationship while investigations continue, and other major labs are scrambling to assess their exposure. The AI industry spent years worrying about what its models might leak; turns out the supply chain was the vulnerability all along.
OpenClaw Gives Users Yet Another Reason to Be Freaked Out About Security
The viral AI agentic tool OpenClaw had a vulnerability that let attackers silently gain unauthenticated admin access — and Anthropic is now charging Claude Code subscribers extra to use it. So you get to pay a premium for the privilege of an attack surface that security researchers are politely suggesting you treat as already compromised. Agentic AI is genuinely powerful and genuinely terrifying, sometimes simultaneously.
Anthropic Launches Cowork, a Claude Desktop Agent That Works in Your Files — No Coding Required
Anthropic’s Cowork brings the autonomous file-wrangling power of Claude Code to non-technical users, and the team apparently built the whole thing in about ten days using Claude Code itself. It’s a genuinely interesting product move — turning a developer tool into something your marketing director might actually use — and the meta-story of “AI built with AI” is either inspiring or a preview of a feedback loop nobody has fully thought through.
AI Agents Promise to ‘Run the Business,’ But Who Is Liable If Things Go Wrong?
A UK financial regulator puts it bluntly: “You can’t blame it on the box.” A global tech analyst adds: good luck finding anyone else to blame either. As AI agents move from demos to actual business operations, the liability question remains a spectacular void — vendors tout the upside and quietly disclaim responsibility for the downside in 6-point font. This is the governance gap that’s going to produce some genuinely ugly case law in the next few years.
Copilot Is ‘For Entertainment Purposes Only,’ According to Microsoft’s Terms of Use
Microsoft is out here charging enterprise customers serious money for a product its own terms of service classify alongside slot machines and novelty toys. To be fair, every AI company buries similar disclaimers in their ToS — it’s just that Microsoft somehow used the most perfectly unhinged phrasing. If your company’s AI strategy is built on “entertainment purposes only” infrastructure, perhaps revisit that roadmap.
Google Bumps Up Q Day Deadline to 2029, Far Sooner Than Previously Thought
Google is now warning the entire industry to migrate off RSA and elliptic curve encryption faster than anyone had planned — the moment when quantum computers can crack today’s encryption is looking like 2029, not some comfortably distant future date. Paired with new research showing quantum attacks require vastly fewer resources than previously assumed, this isn’t a drill anymore. The infrastructure underpinning the entire internet was built for a threat model that’s expiring on a tighter schedule than most organizations have prepared for.
Bottom Line
It’s a Monday where the biggest AI story might not be OpenAI’s $122 billion raise — it’s that between source code leaks, supply chain breaches, unauthenticated exploits, and terms of service disclaimers, the industry is sprinting to deploy agents while the security foundation underneath is actively on fire.