Anthropic’s Claude Fable 5: A Powerful AI Tool Now Released to the Public!

A version of an artificial intelligence (AI) tool, previously deemed too powerful for public release, has hit the market. Anthropic’s Claude Fable 5, an upgrade of the Claude Mythos program, sparked serious alarm among leaders in tech, finance, and government when it was initially rolled out privately back in April for testing. The chatter around this tool has been intense, with some experts expressing concerns that its capabilities could jeopardize financial security. Others, however, are questioning if the buzz is merely a marketing narrative.

On Tuesday, Anthropic announced that Fable will come equipped with safeguards and user restrictions, but they acknowledged, “releasing a model this capable comes with risks.” They emphasized that “Fable’s capabilities exceed those of any model we’ve ever made generally available.” When Mythos was first unveiled to a select group of organizations, Anthropic cited the tool’s intelligence as a potential threat due to its ability to exploit or hack into computer systems.

Canadian Finance Minister François-Philippe Champagne weighed in back in April, stating that the scrutiny surrounding Mythos was justified, partly because it represents “the unknown, unknown.” Meanwhile, despite Anthropic’s ongoing legal battles with the US Department of Defense over the military’s use of its AI tools, it seems that various US government agencies have also been testing Mythos.

Anthropic is on the verge of becoming a publicly traded company, with its private valuation soaring close to $1 trillion (£747 billion). Its consistent demonstration of superior and expansive AI capabilities underscores its allure to investors. The company also revealed that around 150 groups that previously previewed Mythos will now gain access to Claude Mythos 5, which lacks restrictions on cybersecurity or biology, depending on the specific applications of an organization.

So far, organizations and companies using Mythos have reported over 10,000 critical security vulnerabilities in their systems to Anthropic. The firm indicated that this updated access is currently limited to a “small group of cyberdefenders and infrastructure providers,” but anticipates expanding access shortly. “We intend to broaden access to Mythos 5 through a more extensive trusted access program,” the company stated.

Both Fable and Mythos are fundamentally the same model, albeit with different safeguards and access levels. They can operate “unattended” on human commands for longer durations than any prior Claude models, according to Anthropic. Co-founder Jack Clark mentioned last week on BBC Newsnight that the rapid expansion of AI capabilities raises concerns about the pace of technological advancement. “You want the option to be able to take your foot off the gas and put your foot on the brake,” he explained. “Right now, it’s like the AI industry has a gas pedal, but it doesn’t have a brake pedal.”

Kaynak: Orijinal Haber