Anthropic commits 5% of its shares to a new public benefit…
Today's theme is access. Frontier labs are locking in long-term commitments while the websites your agents depend on are closing their doors. If your product needs an agent to reach a page, a login or a checkout, plan for a deal, not a free pass.
- Anthropic's 5% public benefit fund
- Anthropic acquihires Opaque Systems
- Websites closing doors to AI agents
- Quick hits on proxies, bills and bots
- Lounge AI Index and new members
Top stories
All news →What happened: Anthropic is putting 5% of its shares into a new public benefit fund, structured as a wholly owned LLC rather than a 501(c)(3).
- Its equity cannot be used for lobbying or political giving in the US or abroad.
- With co-founder pledges of 80% of their equity and an employee match program, the total dedicated to the public good is roughly 20% of shares.
- The company has already committed $200 million to labor market research and will publish governance and valuation milestone details in coming months.
Why it matters: Frontier labs are formalizing long-term mission commitments ahead of reported IPO plans. Investors and employees will ask how pledges like this affect dilution and governance, so any team drafting its own charter should expect the same questions.
What happened: Anthropic is absorbing confidential computing startup Opaque Systems through an acquihire, and Opaque is winding down operations.
- Opaque began as a UC Berkeley RISELab research project and raised a $24 million Series B at a $300 million valuation earlier this year.
- The startup listed Anthropic among its customers, alongside ServiceNow and Accenture.
- Before the deal, Opaque had shifted from confidential AI to verifiable AI, aiming to produce cryptographic evidence of what an agent ran and under which policies.
Why it matters: Verification and confidential computing are moving into frontier-lab stacks. Founders in those niches should expect talent consolidation, and anyone who relied on a small vendor should review continuity terms.
What happened: Major sites are blocking AI agents from tasks like booking, shopping and account access, often without a clear explanation to users.
- Yelp only allows agents that pay for its data licensing program, Delta Air Lines is blocking ticket bookings, and JPMorgan Chase and Charles Schwab are blocking agents from accounts.
- Cloudflare now blocks AI shopping agents by default on any page that shows an ad, and there is no opt-out directive for agents yet, only a block.
- AI training has grown to 52% of crawler traffic on Cloudflare's network, up from 22% last year, while Meta is partnering with Walmart, Gap, Sephora and Best Buy so Muse can shop their sites.
Why it matters: For a founder, distribution now depends on bilateral deals and data licenses more than on agents loading pages freely. Blocked tasks often fail without a visible cause, so build alerts and fallbacks into your product.
Quick hits
- Anthropic's draft safety bill only triggers at 50 deaths or $1B in damage, and it has no legal force yet. tomshardware.com →
- Inbox Zero Mail is an open-source Superhuman alternative, with AI add-on at $20 per month. Product Hunt →
- A self-hosted LLM proxy turns spend and model choice into data you own, but you must run it. jrdsgl.com →
- MeetPaw adds instant meetings to the Mac menu bar, and its Google Meet links depend on Google's API rules. Product Hunt →
- Meta attributes the No Kings Coalition account suspension to a Claude-linked tool running about four times human speed. TNW →


