Google’s Threat Intelligence Group (GTIG) reported that it has likely thwarted an attempt by hackers to use artificial intelligence for a large-scale exploitation of zero-day vulnerabilities. The hackers were using an AI model, believed to be OpenClaw, to create a method for bypassing two-factor authentication, potentially leading to widespread breaches. Amid growing concerns about AI-facilitated cyber threats, companies like Anthropic have delayed the rollout of their AI models, prompting White House discussions with tech leaders on the associated risks. Meanwhile, hackers linked to nations such as China and North Korea have shown significant interest in exploiting such AI tools for malicious purposes.

Google: Google is a leading technology company that develops AI models like Gemini and provides comprehensive cybersecurity services through its various teams. In this news, Google’s Threat Intelligence Group uncovered and likely prevented a hacker group’s plan to use AI for a mass exploitation event targeting zero-day vulnerabilities. The report highlights Google’s proactive role in countering AI-assisted cyber threats.
Mythos: Mythos is Anthropic’s AI model designed for advanced capabilities but initially held back over fears of criminal exploitation of old software flaws. Its delayed release raised alarms in the industry leading to high-level meetings. It is now available to vetted testers like Apple and Microsoft.
OpenAI: OpenAI develops powerful AI models and is addressing cybersecurity applications. Last week, it announced a limited preview of GPT-5.5-Cyber for vetted cybersecurity teams to counter emerging threats. The news contrasts this defensive use with hackers’ offensive employment of similar tools.
OpenClaw: OpenClaw is an AI model accessible to hackers for identifying software vulnerabilities and developing exploits. The news report cites it as a tool already being used by threat actors to uncover zero-days and launch cyberattacks. Google highlighted its role in ongoing malicious activities by criminal and nation-state groups.
Anthropic: Anthropic is an AI safety-focused company that develops advanced models with safeguards against misuse. It delayed the Mythos model rollout in April due to cybersecurity risks of vulnerability discovery, sparking industry-wide discussions and White House meetings. The model has since been released to select testers including cybersecurity firms.
Microsoft: Microsoft is a major technology company with extensive cybersecurity offerings and AI development. It is testing Anthropic’s Mythos model as part of efforts to safely harness AI against vulnerability exploitation. The news underscores its involvement in industry responses to AI misuse by hackers.
CrowdStrike: CrowdStrike is a leading cybersecurity firm providing endpoint protection and threat intelligence services. It is among the select group testing Anthropic’s Mythos model to evaluate AI safety in defensive contexts. This positions it at the forefront of countering AI-driven cyber threats.
Palo Alto Networks: Palo Alto Networks is a prominent cybersecurity company specializing in network security and threat prevention. It received access to Anthropic’s Mythos model for testing amid concerns over AI-enabled attacks. This reflects its role in collaborative defenses against evolving AI threats.
Google’s Threat Intelligence Group: Google’s Threat Intelligence Group (GTIG) is a specialized team within Google focused on detecting and mitigating advanced cyber threats, including nation-state actors. They issued a report detailing hackers’ use of AI models to exploit zero-day flaws and bypass two-factor authentication, claiming high confidence in thwarting a planned mass attack. GTIG emphasized threats from groups linked to China and North Korea using tools like OpenClaw.

`json
{
“Industry Response”: “Anthropic’s delay of its AI model rollout prompted White House meetings with technology leaders to address cybersecurity risks associated with AI.”,
“Defensive AI Tools”: “OpenAI’s GPT-5.5-Cyber is available in a limited preview for cybersecurity teams to address AI-assisted threats.”,
“AI Threat Evolution”: “Hackers from countries such as China and North Korea are actively using AI for discovering software vulnerabilities and developing malware.”
}
`