Transcript
SPEAKER_1: Pope Leo XIV just dropped forty-three thousand words on AI. That is the first papal encyclical focused entirely on artificial intelligence, and it's called Magnifica humanitas. Walk me through what's actually in it. SPEAKER_2: It calls for AI regulation, child protection from hypersexualized AI imagery, and urges governments to slow down AI development. It also decries what it calls new forms of slavery in AI and tech supply chains. It's a sweeping moral document, not a technical one. SPEAKER_1: And Anthropic's fingerprints are reportedly all over it. What's the connection there? SPEAKER_2: Anthropic co-founder Christopher Olah attended the unveiling in person. The Washington Post reported that a section on AI's unpredictability suggests Anthropic's influence. Religion News Service documented months of ongoing ethics discussions between Anthropic and the Vatican that led to that invitation. SPEAKER_1: And in that same week, Anthropic is finalizing a classified NSA contract. So they're briefing the Pope and the spy agencies in the same news cycle. SPEAKER_2: Exactly. That's a genuinely new posture for any company. And it wasn't just Anthropic at the Vatican either. Meta, Google, and Amazon execs met Vatican officials on April twenty-ninth as part of a quiet lobbying push ahead of the encyclical's release, per Politico. SPEAKER_1: Tech is now lobbying moral authorities with the same urgency it brings to capitols. That's a real shift. Let's move to Washington, because it was a massive week for AI spending there too. SPEAKER_2: Washington committed more than eleven billion dollars to AI in just seven days. Two billion in equity stakes in quantum companies, nine billion in chips for spy agencies, and a Pentagon task force to study how to safely deploy AI tools with hacking capabilities across Cyber Command and NSA missions. SPEAKER_1: And at the same time, Trump killed his own AI executive order. What happened there? SPEAKER_2: The White House postponed the signing ceremony. Trump said he didn't like certain aspects of it. Sources told the Washington Post he shelved it after calls with Musk and Zuckerberg. Politico reports David Sacks made the case that federal AI reviews would slow innovation and hurt the US in the race against China. SPEAKER_1: And Axios's read on it was simply that Trump just hates regulation. SPEAKER_2: That was their take, yes. The draft order had emphasized that government AI reviews would be voluntary, and Treasury's expanded role was flagged as a final-hour concern. Either way, the rules died and the checks cleared. Less regulation, more procurement. SPEAKER_1: On the quantum side, the Commerce Department announced two billion in grants to nine companies and the government is actually taking equity stakes. IBM gets one billion of that package. Markets reacted immediately. SPEAKER_2: Very immediately. D-Wave closed up thirty-three percent, Rigetti up thirty percent, IBM up twelve percent. And France matched the bet hours later, with Macron announcing one billion euros for quantum research and five hundred fifty million for semiconductors. SPEAKER_1: Industrial policy with sovereign-fund mechanics. Now let's get into the cybersecurity story, because this was a huge week on that front too. Five thousand five hundred plus GitHub repos infected in one coordinated attack. SPEAKER_2: This is being called the Megalodon supply-chain attack. More than five thousand five hundred GitHub repositories were infected with malware on May eighteenth, relying on automated commits. GitHub also linked a separate breach of three thousand eight hundred internal repositories to the TanStack npm supply-chain attack, executed via a malicious Nx Console VS Code extension. SPEAKER_1: And the same crew behind the GitHub breach ran even more attacks on top of that? SPEAKER_2: Yes. Wired reports TeamPCP, the gang behind the GitHub breach, also ran twenty waves of supply-chain attacks recently, compromising five hundred plus pieces of software. It's a coordinated, scaled operation. SPEAKER_1: And on the defensive side, Anthropic's Project Glasswing has identified ten thousand plus high- or critical-severity vulnerabilities using Claude Mythos Preview since launch. But there was also a troubling flip side this week. SPEAKER_2: Right. The NTSB suspended its civil accident database after people used AI to re-create voices of pilots killed in a twenty twenty-five UPS crash. Even authoritative public datasets are now an attack surface. Both sides of the security war now have AI doing the heavy lifting. SPEAKER_1: There was also a big arrest this week. A twenty-three-year-old named Jacob Butler, alleged operator of the Kimwolf DDoS botnet, which infected around two million devices. And security engineer job postings are up eleven percent year over year in Q one. SPEAKER_2: Driven by threats from AI-generated code and models like Mythos and GPT-5.4-Cyber. The hiring signal tells you where the pressure is being felt. SPEAKER_1: Let's run through the funding and M and A highlights. There's a lot here. Starting with Manus, which is in talks to raise over one billion dollars to buy itself back from Meta. SPEAKER_2: Beijing forced the unwind of Meta's two billion dollar acquisition, so Manus is now raising to reclaim its independence. Meanwhile DeepSeek has a mid-ten-billion-dollar round in progress, with founders pitching an AGI focus over short-term commercialization to potential investors. SPEAKER_1: And Hark closed a massive Series A. Seven hundred million plus at a six billion dollar valuation. Brett Adcock's new AI hardware devices company, led by Parkway Venture. SPEAKER_2: Then Modal Labs raised three hundred fifty-five million at four point six five billion, up from one point one billion in twenty twenty-five. That's a serverless cloud platform for AI apps and inference. Exa raised two hundred fifty million at two point two billion, up from seven hundred million in September, for a search engine built for AI agents. A sixteen z led that one. SPEAKER_1: Mercury raised two hundred million Series D at five point two billion. TCV led. And there were several others: Farther Finance at one hundred fifty million, Fresha at eighty million from KKR, Moment at seventy-eight million, Commure at seventy million, Socket at sixty million, and Catena Labs at thirty million. SPEAKER_2: Catena Labs is notable because it's applying for a US bank charter. It's building AI-agent financial guardrails, and it was founded by Circle co-founder Sean Neville. That's a company to watch. SPEAKER_1: On the IPO side, OpenAI is preparing to file confidentially as soon as May twenty-second, targeting a public debut as early as September. And two others filed confidentially this week as well. SPEAKER_2: Oura, the smart ring maker last valued at eleven billion in September twenty twenty-five, filed confidentially for a US IPO set for later in twenty twenty-six. And Blockchain dot com, the UK-based crypto exchange founded in twenty eleven and once valued at fourteen billion, also filed confidentially. SPEAKER_1: There were also notable M and A moves. Uber made a takeover offer for Delivery Hero at thirty-eight euros per share, valuing the group at over eleven point five billion euros. That offer was rebuffed and Uber is weighing a higher bid. SPEAKER_2: And the Anthropic, Blackstone, and H and F joint venture made its first deal, acquiring Fractional AI. Sources say Fractional ended its OpenAI deal as part of the acquisition. Plus Tether bought out SoftBank's twenty-six percent stake in bitcoin treasury firm Twenty One Capital for six hundred seventy-nine million, taking Tether to roughly seventy-one percent ownership. SPEAKER_1: And Activision shareholders reached a two hundred fifty million dollar settlement with Microsoft over allegations they were underpaid during the twenty twenty-three acquisition. SPEAKER_2: A lot of loose ends getting tied up, and a lot of new ones being created. SPEAKER_1: So let's bring it all together. The thread connecting everything this week is institutional power finding its voice on AI. The Pope released forty-three thousand words on AI ethics. Washington committed eleven billion plus while killing its own regulatory order. Five thousand five hundred repos got hit in one attack while Glasswing found ten thousand critical vulns. The state isn't asking permission anymore. It's writing checks and buying capability.