Anthropic announced this Tuesday (2) the expansion of access to its artificial intelligence (AI) model Mythos for over 150 organizations distributed in more than 15 countries. Expansion occurs through Project Glasswingan initiative created to evaluate the use of AI in identifying software vulnerabilities and strengthening cybersecurity practices.
According to the company, new participants include sectors that had limited representation in the initial phase of the program, such as energy, water supply, healthcare, communications and hardware. Before being granted access to the model, organizations will need to meet security requirements defined by Anthropic.
Project expands reach to new sectors
In publication in blog of the company, Anthropic stated that the expansion represents another step in its long-term objectives related to digital security.
According to the company, the goal is to use artificial intelligence to make software more secure and help the sector adapt to the changes that technology can cause in fundamental assumptions of the company. cybersecurity.
The expansion of Project Glasswing was announced one day after Anthropic reported that it will begin offering access to Mythos on European Union. On Monday, the company also announced that it confidentially submitted its prospectus for an initial public offering (IPO) to the Securities and Exchange Commission (SEC), advancing in this process before rival OpenAI.
Mythos capabilities sparked debate
The first test group of Mythos was launched in April to 50 partners. Since then, the model has attracted attention for its ability to locate faults in systems and software.
Experts have expressed concern that advanced artificial intelligence could make the work of hackers easier, allowing for faster identification of vulnerabilities. While many have highlighted that these capabilities already existed previously, Mythos could accelerate this process.
As concerns grew, the White House held several meetings with representatives from large technology companies and financial institutions to discuss ways to safely implement advanced AI models.
More than 10 thousand failures identified
According to Anthropic, partners participating in Project Glasswing have identified more than 10 thousand vulnerabilities classified as high or critical level since the launch of the initiative.
The company estimates that a major cyberattack could affect more than 100 million people.
Among the main members of the program are Apple, Nvidia, Microsoft, CrowdStrike and Palo Alto Networks. Anthropic did not reveal which other companies will join the coalition.
The cloud data management platform Rubrik said in a statement that it is among the organizations included in this new stage of the program.
Source: www.olhardigital.com.br
Source link
