Australia’s nationwide safety businesses and significant infrastructure defenders are being granted entry to Anthropic’s Claude Mythos Preview AI mannequin, which the corporate has not launched publicly because of its superior cybersecurity capabilities.
Anthropic introduced this week that it will increase entry of Mythos Preview to organisations in 15 new nations beneath its so-called Mission Glasswing.
A spokesperson for the Australian Alerts Directorate (ASD) mentioned the intelligence company welcomed Anthropic’s enlargement of the challenge to incorporate the Australian authorities and Australian personal corporations.
An Anthropic spokesperson confirmed Australian organisations would acquire entry to Mythos Preview, however didn’t identify any particular organisations or disclose the opposite new nations added to Mission Glasswing.
Anthropic had beforehand solely offered the AI mannequin to the governments of the US and United Kingdom, in addition to round 50 primarily American corporations and organisations.
That tally will now attain round 200 organisations globally.
The brand new additions mirrored “a number of industries that weren’t properly represented in our preliminary cohort, akin to energy, water, healthcare, communications, and {hardware}”, Anthropic mentioned.
“And most of the new companions are distributors—corporations or nonprofits that preserve codebases which are relied upon by numerous different organisations around the globe, together with governments.”
Different nations which is able to acquire entry to Mythos Preview reportedly embody France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium, Sweden, India, Japan, and South Korea, in keeping with the Financial Times.
The enlargement of Mission Glasswing may also reportedly see the NATO navy alliance, the European Union’s cybersecurity company ENISA, and a few monetary market operators acquire entry to Mythos Preview.
Anthropic publicly introduced Mythos Preview in April after the mannequin supposedly found thousands of “high-severity vulnerabilities”, together with some in each main working system and internet browser.
Having shared the know-how with main corporations akin to Apple, Google, Microsoft, AWS, and Nvidia, Anthropic mentioned Mythos Preview had since discovered “greater than 10,000 high- or critical-severity safety flaws”.
Australian cybersecurity officers and monetary regulators warned local organisations to bolster their security posture within the weeks after Mythos Preview’s announcement, when Anthropic reportedly briefed authorities businesses with out granting entry to the mannequin.
Trump indicators government order on AI testing
Safety considerations raised by Mythos Preview and superior AI fashions from different companies akin to OpenAI contributed to US President Donald Trump’s choice to signal a brand new government order on Tuesday.
The order asks American companies to voluntary share their strongest AI fashions for presidency cybersecurity testing earlier than they’re launched publicly.
It directs authorities businesses to achieve agreements with AI builders to check their fashions as much as 30 days earlier than any wider launch – a shorter interval than the 90-day interval reportedly listed in an order which Trump shelved in late Could.
The president postponed plans to signal the earlier order over considerations that the measure might hurt America’s progress in AI.
“We’re main China, we’re main everyone, and I don’t wish to do something that’s going to get in the way in which of that lead,” Trump advised reporters on the time.
Anthropic, which has beforehand clashed with the Trump administration over authorities entry to its AI fashions, described the president’s new government order as “an vital step in strengthening America’s management in AI”.
“We look ahead to collaborating with the White Home to assist its implementation,” the corporate mentioned.
OpenAI CEO Sam Altman mentioned on Wednesday that Trump’s new government order “will get the steadiness proper”.
“The US ought to lead on AI by persevering with to develop the easiest fashions, ensuring they’re secure, and getting cyber instruments into the fingers of trusted defenders,” he mentioned.
Anthropic information IPO forward of OpenAI
Information of Anthropic’s enlargement of Mission Glasswing got here a day after the corporate confidentially filed for an preliminary public providing (IPO), in a race to beat ChatGPT maker OpenAI to the general public market.
Whereas Anthropic didn’t disclose the main points of its IPO, the corporate’s newest funding spherical noticed it increase $US65 billion ($90.6 billion) at a valuation of round $US965 billion ($1.35 trillion).
After fending off a recent legal challenge from Elon Musk, OpenAI is reportedly getting ready to file an IPO of its personal, which might worth the agency at an analogous stage to Anthropic.
Musk’s personal SpaceX, which acquired xAI earlier this yr, detailed its own IPO in Could with a goal valuation of round $US1.75 trillion ($2.4 trillion).
