The announcement on Aug. 24 paired a checkbook with a security update. Anthropic said it is launching a $35 million fund to support open-source AI development, and it expanded access to Mythos 5, its model for defensive security teams, which the company calls Defenders. The two moves, taken together, show how Anthropic is positioning itself in the months before a planned public listing: generous to the ecosystem, strict about safety, and eager to be seen as both.
The open-source fund is a calculated bet. Anthropic has built its brand on safety and on a business model that sells access to its proprietary models, and open source has sometimes seemed like the opposite of its strategy. The fund changes the optics, putting real money behind open-source projects and inviting the community to see the company as an ally rather than a gatekeeper. The amount, $35 million, is small relative to the company’s ambitions but large enough to fund hundreds of projects.
The Mythos 5 expansion is the other side of the message. Mythos, Anthropic’s line of models for security professionals, is designed to help defenders identify and respond to threats, and the company has been expanding who can use it. The announcement extends access to more defensive teams, a move that reinforces the company’s safety narrative while also building relationships with the security professionals who influence enterprise buying decisions.
The timing is deliberate. Anthropic is preparing for an IPO, and its roadshow will emphasize two things: growth and responsibility. The open-source fund gives the company a story about the ecosystem, and the Mythos expansion gives it a story about defense. Together, they position Anthropic as the AI company that cares about both, a message designed for the institutional investors who will underwrite its listing.
The moves also respond to competitive pressure. OpenAI, Anthropic’s chief rival, has built its own open-source and safety programs, and Google has been expanding its security offerings. Anthropic’s announcements are, in part, a way of claiming territory: the company that leads on safety, and the company that supports the open ecosystem, before its competitors can make the same claims.
The open-source community has greeted the fund with a mix of enthusiasm and skepticism. Enthusiasts see a large company putting real resources behind projects that have long been underfunded; skeptics note that corporate support for open source often comes with strings attached, and that the fund’s governance and its relationship to Anthropic’s proprietary interests will determine how much it is worth. The company has said the fund will be administered with community input, a promise that will be tested in practice.
The security community is watching the Mythos expansion for similar reasons. Broader access means more defenders with powerful tools, but it also means more scrutiny of the tools themselves, their limitations, and the claims made about them. Anthropic has said it will publish evaluations of Mythos’s capabilities, and the quality of those evaluations will shape the product’s reputation.
For investors, the announcements are a preview of the company’s public posture. Anthropic is telling the market that it will be a good citizen of the AI ecosystem, that it takes safety seriously, and that it sees both as compatible with growth. The $35 million fund and the Mythos expansion are small line items in the company’s financials, but they are large statements about the kind of company it intends to be.
The fund’s structure, according to the announcement, is designed to be flexible: it will support projects across the AI stack, from models and tooling to evaluation and safety research, and it will be open to applications from independent developers, academic labs, and small companies. Anthropic has said it will prioritize projects that demonstrate real user adoption and that contribute to the ecosystem rather than duplicating existing work. The details will matter, because open-source communities have learned to be wary of corporate funds that arrive with conditions attached, and the fund’s credibility will depend on its independence from Anthropic’s commercial interests.
The Mythos expansion, similarly, is as much about positioning as about product. Security teams have become influential buyers in the AI market, and the companies that win their trust early are likely to keep it through the procurement cycles to come. By expanding access to Mythos 5, Anthropic is building relationships with the people who will recommend, or reject, its products inside enterprises. The strategy is long-term, and it is consistent with the company’s broader approach: invest in trust now, convert it into revenue later, and tell the story of both to the investors who will own the stock.
The test will come after the listing, when the promises made in private to investors meet the public record. For now, Anthropic is doing what every company does before an IPO: telling the best version of its story. The fund and the expanded access are the latest chapters, and they are written to be read together.
