Anthropic Hit by Back-to-Back Leaks
Anthropic is back in the spotlight for the second time in just a few days. Last week, details about their new model Mythos they called “the best they’ve ever built” were accidentally published. Now, part of Claude Code’s source code was unintentionally made accessible to the public. The leak was caused by a mistake in an update by a developer and was spotted by this X user. Soon after, a GitHub repo publishing the captured source code under the name Claw Code hit a record number of stars in less than 12 hours. We could say Claude Code became open source by accident.
While the reason behind the model leak looked like a mistakenly published blog post, this source code leak raised far more serious security concerns. For Anthropic, which has also been flagged by the U.S. Department of Defense over risk concerns with calls to halt its use in government, this might translate into a global reputational hit.
After two major operational mistakes in such a short window, Anthropic has not yet issued a written statement.
OpenAI Raises a Massive Amount of Money After Major Product Resets
OpenAI announced that it has closed a $122 billion funding round, pushing the company’s post-money valuation to $852 billion. In its statement, OpenAI said the capital will be used to support “the next phase”. With $3 billion of that money reportedly coming from individual investors via banks, this can reasonably be read as a warm-up toward a public offering.
After shutting down Sora last week and also announcing an indefinite pause on its “adult content mode” plans, OpenAI seems to be moving away from its experimental “we’re trying everything” mode and adopting a more institutional profile that reassures investors. Taken together, these moves may signal a strategic effort to shed the dead weight and project tighter operational discipline on their way to IPO.
