How fast can AI-powered attackers exploit vulnerabilities?

In one incident, AI exploited 50 different vulnerabilities and attack paths in 10 hours. That would normally take two weeks for human actors.

What evidence supports the warning?

Palo Alto Networks participated in Anthropic's Project Glasswing and OpenAI's Daybreak program. They discovered security vulnerabilities that would normally take a year to unearth.

AI Safety & AlignmentTop Companies' AI MovesTop Companies AI7h agoAug 29, 2026, 06:30 JST1 min read

Unit 42: AI tips cyber balance to attackers

Send on LINE

Key takeaway

AI is shifting cyber balance toward attackers. A lone actor can now launch sophisticated attacks.
Palo Alto Networks found vulnerabilities a year's work in months.
This wave is arriving now, researchers warn.

3 Key Points

What happened
Palo Alto Networks' Unit 42 researchers warn that AI is enabling cyberattacks faster than defenses can handle. After months in Anthropic's Project Glasswing and OpenAI's Daybreak program, they found vulnerabilities that would normally take a year to uncover.
Why it matters
A single threat actor with such AI can carry out nation-state-level attacks without extensive training, says Sherrod DeGrippo of Unit 42. Researchers are investigating an incident where AI exploited 50 vulnerabilities in 10 hours—a task that usually takes humans two weeks.
What to watch
In May, PAN predicted a three-to-five-month window before adversaries exploit vulnerabilities at unprecedented speeds. Now, five months later, Sam Rubin says the wave is arriving. PAN also signed an open letter with 100 companies urging action against weaponized AI.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Get news like this every morning — free

Context & Analysis

The article reports a generational shift in cybersecurity's balance of power, driven by advanced AI. Palo Alto Networks' researchers, through controlled access to frontier AI programs, gained insight into how quickly attackers could work. Their May warning of a three-to-five-month window before unprecedented exploitation speeds now appears to be materializing, as noted by Sam Rubin.

The stakes are underscored by a specific incident where AI enabled an attacker to achieve in 10 hours what would typically take two weeks. This efficiency suggests malicious actors with these tools can operate at a scale once reserved for well-resourced groups. The industry response includes a collective call for action, with 100 companies signing an open letter, indicating a recognized urgency to address the threat of weaponized AI.

FAQ

How fast can AI-powered attackers exploit vulnerabilities?: In one incident, AI exploited 50 different vulnerabilities and attack paths in 10 hours. That would normally take two weeks for human actors.
What evidence supports the warning?: Palo Alto Networks participated in Anthropic's Project Glasswing and OpenAI's Daybreak program. They discovered security vulnerabilities that would normally take a year to unearth.