Google Cloud

GTIG AI Threat Tracker: From Prompting to Autonomy  The Evolution of Adversarial AI

Chinese hackers are running AI on stolen networks to avoid detection, Google says

The Chosun Daily

AI Cyberattacks Steal Thousands in 6 Hours

Gizmodo

Silicon Valley’s AI Agent Push Has Been Paying Off—for Cybercriminals

Belfer Center

Deception and Detection: Why Artificial Intelligence Empowers Cyber Defense over Offense

Openai

The Defenders Window

Infosecurity Magazine

Agentic AI will Supercharge Cyber Threats. But Not the Way You Thin

Center for a New American Security

Tipping the Scales

US Department of War

Csa_China_Based_Ai_Companies_Malicious_Distillation_Against_Us.Pdf

Metaculus

Will cyberattack targeting AI cause power blackout in the US before 2028?

Google's Threat Intelligence Group said Tuesday that hackers have shifted from basic AI prompting to agentic workflows, in one case compromising a cloud resource and then planning, building and executing a mass credential harvesting campaign in under six hours.

Google's Threat Intelligence Group said Tuesday that hackers have shifted from basic AI prompting to agentic workflows, in one case compromising a cloud resource and then planning, building and executing a mass credential harvesting campaign in under six hours.

Google said a group it tracks as UNC6508, linked to China, has targeted academic, medical and military research institutions in North America since 2023, focusing on proprietary AI research. Google did not identify any of the victims.

Google said a group it tracks as UNC6508, linked to China, has targeted academic, medical and military research institutions in North America since 2023, focusing on proprietary AI research. Google did not identify any of the victims.

According to the report, UNC6508 deployed open-source AI models inside compromised cloud environments, which lets attackers use victims' computing resources while avoiding the monitoring built into commercial AI application programming interfaces.

Ynetglobal

According to the report, UNC6508 deployed open-source AI models inside compromised cloud environments, which lets attackers use victims' computing resources while avoiding the monitoring built into commercial AI application programming interfaces.

Liu Chang, a spokesperson for the Chinese Embassy in Washington, rejected the allegations, saying China "opposes hacking activities and fights such activities in accordance with the law" while rejecting what he called smears under the pretext of cybersecurity.

Liu Chang, a spokesperson for the Chinese Embassy in Washington, rejected the allegations, saying China "opposes hacking activities and fights such activities in accordance with the law" while rejecting what he called smears under the pretext of cybersecurity.

The report said a group tracked as UNC6780, also called TeamPCP, embedded prompts referencing biological and nuclear weapons in malware loader scripts, aiming to make AI security scanners refuse or skip analysis of the malicious code beneath.

The report said a group tracked as UNC6780, also called TeamPCP, embedded prompts referencing biological and nuclear weapons in malware loader scripts, aiming to make AI security scanners refuse or skip analysis of the malicious code beneath.

The National Security Agency, CISA and the FBI published a joint advisory Sept. 8 accusing six China-based AI firms, including DeepSeek, Moonshot AI and Alibaba, of extracting billions of tokens from U.S. frontier models such as Claude, GPT, Gemini and Grok.

International Business Times UK

The National Security Agency, CISA and the FBI published a joint advisory Sept. 8 accusing six China-based AI firms, including DeepSeek, Moonshot AI and Alibaba, of extracting billions of tokens from U.S. frontier models such as Claude, GPT, Gemini and Grok.

Google Cloud #

NBC #3261

The Chosun Daily #

Gizmodo #

Belfer Center #

Openai #

Infosecurity Magazine #

Center for a New American Security #

US Department of War #

Metaculus #

Google's Threat Intelligence Group said Tuesday that hackers shifted from basic AI prompting to agentic systems that automate wide swaths of their intrusions. The report said some AI-enabled campaigns can now be conducted in less than six hours.

Google's Threat Intelligence Group said Tuesday that hackers shifted from basic AI prompting to agentic systems that automate wide swaths of their intrusions. The report said some AI-enabled campaigns can now be conducted in less than six hours.

U.S. agencies said Tuesday that six China-based AI companies extracted billions of tokens from frontier models including Claude, GPT, Gemini and Grok. Google also reported regular coordinated distillation campaigns against its proprietary AI systems.

U.S. agencies said Tuesday that six China-based AI companies extracted billions of tokens from frontier models including Claude, GPT, Gemini and Grok. Google also reported regular coordinated distillation campaigns against its proprietary AI systems.

Google reported Tuesday that China-linked group UNC6508 targeted academic, medical and defense institutions in North America for proprietary AI work. The group deployed open-source models on compromised cloud systems to avoid commercial monitoring.

Google reported Tuesday that China-linked group UNC6508 targeted academic, medical and defense institutions in North America for proprietary AI work. The group deployed open-source models on compromised cloud systems to avoid commercial monitoring.

Google said Tuesday that Iranian actors used Gemini to generate photorealistic personas by specifying camera angles, studio lighting and facial texture. The operators used detailed prompts rather than adjusting images through repeated manual requests.

Google said Tuesday that Iranian actors used Gemini to generate photorealistic personas by specifying camera angles, studio lighting and facial texture. The operators used detailed prompts rather than adjusting images through repeated manual requests.

Google said Tuesday that North Korea-linked group Midnight Neptune used AI to create malware aimed at companies and to steal cryptocurrency. Google's quarterly report described growing AI use across financially motivated and state-linked groups.

Google said Tuesday that North Korea-linked group Midnight Neptune used AI to create malware aimed at companies and to steal cryptocurrency. Google's quarterly report described growing AI use across financially motivated and state-linked groups.

The White House has described the United States and China as racing to develop cutting-edge AI. U.S. agencies said Tuesday that industrial-scale distillation can reduce the research and computing resources needed to match U.S. frontier models.

The White House has described the United States and China as racing to develop cutting-edge AI. U.S. agencies said Tuesday that industrial-scale distillation can reduce the research and computing resources needed to match U.S. frontier models.

Google: Hackers Automate Attacks With Agentic AI

Google: Hackers Automate Attacks With Agentic AI

Go Deeper

Sign Up for Our Free Newsletters
Sign Up for Our Free Newsletters

Google: Hackers Automate Attacks With Agentic AI

Google: Hackers Automate Attacks With Agentic AI

Go Deeper

Sign Up for Our Free NewslettersSign Up for Our Free Newsletters

Sign Up for Our Free Newsletters
Sign Up for Our Free Newsletters