Anthropic, the artificial intelligence company behind the popular chatbot Claude, is experiencing one of the most intense periods in its short history . Between preparations for its highly anticipated IPO, advancements in its Mythos cybersecurity model, and growing pressure from its own employees to slow the pace of development in the sector, the company has become the center of debate about the future of AI.
While investors closely monitor the company's movements, its CEO, Dario Amodei, has had to grapple with difficult decisions, ranging from withholding potentially dangerous technologies to positioning itself in the regulatory debate. All of this is happening against a backdrop of the global IPO market attempting to regain the momentum it lost in 2021.
Anthropic prepares for its stock market debut
According to various industry sources, Anthropic plans to go public later this year , possibly in the last quarter. The IPO comes amid a global surge in listings: over €173.000 billion was raised worldwide in the first half of 2026, although this figure still falls short of the record set in 2021. The company, along with OpenAI, is one of the biggest hopes for the market's recovery.
Anthropic's valuation is around one trillion dollars, which would make it one of the biggest IPOs of the year . However, geopolitical uncertainty and the US midterm elections could complicate the plans. In Europe, the volume of IPOs remains modest compared to the US market, which accounts for the majority of fundraising.
Claude Mythos: the model that breaks cryptographic algorithms
One of Anthropic's most striking developments is Claude Mythos Preview, an artificial intelligence model capable of autonomously finding vulnerabilities in cryptographic algorithms . In recent tests, Mythos managed to improve upon the best-known attack against the HAWK post-quantum digital signature scheme in just 60 hours, halving the effective strength of its keys. It also devised an improved technique for attacking a reduced version of AES encryption, between 200 and 800 times faster than previous methods.
These results led Anthropic to decide against openly releasing the model , since, in Amodei's words, "Mythos is a total weapon" and should require a license for its use. The company coordinated the disclosure with NIST and notified the creators of HAWK before making the findings public. The cost of each attack was around $100.000 in APIs, and researchers from ETH Zurich and the universities of Tel Aviv and Haifa participated in its development.
Anthropic employees join the call to slow down AI
More than 1,000 employees from leading artificial intelligence labs, including Anthropic, OpenAI, and Google DeepMind, have signed an open letter to the U.S. government requesting its support for an international mechanism to "deliberately pace" AI development. The petition, titled "Pacing the Frontier," is endorsed by Dario Amodei himself and Jared Kaplan, co-founder and chief scientist at Anthropic.
The signatories warn that the automation of AI research could lead to a loss of control over the technology if measures are not taken in time. They propose creating governance and oversight tools that would allow for slowing down or temporarily pausing the most dangerous advances. Although OpenAI's executive director, Sam Altman, did not sign the letter, he has acknowledged in interviews that moderating the pace may be necessary, though he fears this could be perceived as regulatory capture.
Anthropic thus finds itself at a crossroads: its ambition to go public and compete in the AI ​​market clashes with the responsibility of managing dual-use technologies and with pressure from its own employees to ensure the sector doesn't advance faster than society can absorb. Between the IPO, Mythos, and calls for a slowdown, Dario Amodei's company is setting the agenda for a debate that will define the future of artificial intelligence.
