
Google projects autonomous AI cyberattacks within three to six months
Google and Mandiant forecast autonomous agentic attacks within six months as Palo Alto Networks reports on a $21.2 billion backlog and $3.3 billion quarterly revenue guidance.
- By
- Sophie Lindqvist
- Filed
- Channel
- AI & Compute
- Read
- 4 min read
By Sophie Lindqvist
Google Threat Intelligence Group and Mandiant project sophisticated, fully autonomous agentic attacks within three to six months, according to an AI threat tracker published Nov. 16, 2026.
The forecast is a roadmap, not proof that such an attack has already occurred. The tracker says current AI tools can run several attack steps with minimal human involvement.
Palo Alto Networks (NASDAQ: PANW) reports fiscal first-quarter 2027 results after the close on Wednesday, Nov. 18. The results arrive amid a fresh threat forecast from a major rival and a Google product announcement that overlaps Palo Alto’s security market.
What has Google actually documented?
Google describes current attacker activity as AI-assisted operations with little human involvement across several steps. It does not claim that current agents independently complete every stage.
Its two later stages remain projections:
- Multi-agent, simple attacks with little or no human interaction within one to three months.
- Fully autonomous, sophisticated agentic attacks within three to six months.
Threat actors used four named open-source AI penetration-testing frameworks:
- ARTEX.
- CAIRN, which targeted the U.S. aerospace sector.
- CyberStrikeAI.
- PentAGI.
An exposed staging server held a system called exploit_agent. It compared code before and after patches, wrote exploits and ran them. Google says the system did so “without a human in the loop for the first time.”
The exploits targeted flaws that already had patches. Google therefore identifies patching speed as the main defense and says the trend favors automated security platforms. It offers no revenue estimate for that benefit.
Why security budgets may expand?
Google reports that vulnerability disclosures and exploits nearly doubled between January and August 2026.
Zero-day exploits target flaws before patches exist. They averaged eight per month in 2025 and 11 per month in 2026, then reached 22 in August.
In LLM-jacking cases, attackers take over corporate cloud and AI accounts. Google says single-day fraudulent charges exceeded $100,000 in those incidents.
The tools also sacrifice stealth for speed and leave tracks. That behavior may support detection products such as Palo Alto Networks’ Cortex and Unit 42. The tracker provides no Palo Alto attribution, customer count or revenue figure.
How directly does Google compete?
Also on Nov. 16, Google promoted Google AI Threat Defense. The package combines Gemini with Wiz, Google’s CodeMender patching agent, and Mandiant.
Palo Alto’s filing already identifies Alphabet as a competitor that builds security into its products. Google thus supplies both the threat forecast and a bundled response.
The source places Alphabet’s value at about $3.49 trillion and says Google Cloud revenue grew 103% last quarter. It provides no comparison period or revenue base, limiting the growth claim.
Security is Palo Alto’s entire business. Its Next-Generation Security annual recurring revenue rose 63% to $9.10 billion. Google provides no stand-alone revenue or margin figure for its security package.
What has the market priced?
Palo Alto’s fiscal 2027 revenue guidance calls for $14.10 billion to $14.20 billion. Remaining performance obligations, meaning contracted revenue not yet booked, reached $21.2 billion.
The source says the market now values that backlog below its October level. It offers no dollar comparison to quantify that valuation change.
Palo Alto beat estimates in seven straight reported periods. Yet its shares averaged a 4.53% decline on report days, a record that makes an early post-earnings drop consistent with its history.
Nov. 17 trading:
- Palo Alto fell 2.15% to $309.29, its fourth straight decline and lowest close since June 26, 2026. It closed 26.3% below its $419.91 Oct. 6 peak. An after-hours quote was $315.49 at 7:59 p.m. ET.
- Alphabet slipped 1.24% to $280.20.
- Fortinet rose 1.00%.
- CrowdStrike fell 1.63%.
- Zscaler fell 2.96%.
No source ties those moves to Google’s tracker. The article’s point is narrower: Google can forecast the threat and sell defenses against it.
The source describes Palo Alto as the group’s most broadly exposed vendor because it covers network, cloud, identity and security operations. Those layers match the chain that multi-agent attacks traverse. Breadth may aid cross-selling, but no conversion data supports that claim.
What must the earnings call establish?
Investors will focus on two tests identified in the source:
- Management addresses AI-driven attacks and their effect on security demand.
- Quarterly revenue lands within the $3.300 billion to $3.310 billion guidance range.
Revenue inside that range would support the argument that AI threats are lifting demand; revenue below it would weaken that case. The source provides no analyst consensus, so investors cannot use the article to measure the size of any beat or miss.
The next commercial test is whether Palo Alto converts rising exploit volume into paid demand before Google bundles AI, patching and threat-intelligence functions across its own stack. The tracker establishes urgency, not a revenue forecast.
Source: 247wallst.com
More from Sophie Lindqvist
Show full bio
News editor covering money, funding and market shifts around Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,153 articles
Related articles
ai-agents-ran-25-intrusions-against-chip-and-defense-targets
AI agents ran 25 intrusions against chip and defense targets
fbi-warns-ai-hacking-tool-artex-breaches-networks-in-under-three-hours
FBI Warns AI Hacking Tool Artex Breaches Networks in Under Three Hours
crowdstrike-ai-targeted-200-plus-organizations-across-14-countries
CrowdStrike: AI targeted 200-plus organizations across 14 countries
google-says-ai-now-builds-and-runs-exploits-with-no-human-involved
Google Says AI Now Builds and Runs Exploits With No Human Involved



