This Week in AI Security - 15th January 2026

Happy New Year! Jeremy kicks off 2026 with a special extended episode to catch up on everything that happened while the industry was on holiday. From humanoid robots to new global protocols for "Agentic Commerce," AI adoption is accelerating at an unprecedented pace.

Market & Strategic Trends:

• Explosive Growth: AI consumption has tripled over the last year, with user prompt volume growing 6x.
• Specialized Foundations: We are seeing a shift from general-purpose models to domain-specific LLMs, such as Nvidia's Alpamayo for autonomous vehicles.
• Agentic Commerce: Google has announced a new protocol designed to facilitate interactions between AI shopping agents and retail systems.
• Regulatory Landscape: New York has introduced the RAISE Act for AI security, while Italy is challenging Meta's "walled garden" approach to AI chatbots on WhatsApp.

Critical Vulnerabilities & Research:

• Prompt Injection is "Inherent": OpenAI researchers suggest that agentic browsers may be inherently vulnerable to indirect prompt injection due to their need to process external instructions.
• Supply Chain Risks: Major vulnerabilities were identified in LangChain (API serialization issues) and n8n (max severity RCE), both core tools for building AI workflows.
• Shadow AI Attacks: Over 91,000 attack sessions were detected targeting AI deployments, including Server-Side Request Forgery (SSRF) campaigns launched via Llama.

Episode Links

https://securityboulevard.com/2026/01/report-increase-usage-of-generative-ai-services-creates-cybersecurity-challenge/

https://techcrunch.com/2026/01/05/boston-dynamicss-next-gen-humanoid-robot-will-have-google-deepmind-dna/

https://techcrunch.com/2026/01/05/nvidia-launches-alpamayo-open-ai-models-that-allow-autonomous-vehicles-to-think-like-a-human/

https://techcrunch.com/2026/01/11/google-announces-a-new-protocol-to-facilitate-commerce-using-ai-agents/

https://techcrunch.com/2025/12/20/new-york-governor-kathy-hochul-signs-raise-act-to-regulate-ai-safety/

https://techcrunch.com/2025/12/24/italy-tells-meta-to-suspend-its-policy-that-bans-rival-ai-chatbots-from-whatsapp/https://github.com/asgeirtj/system_prompts_leaks/

https://techcrunch.com/2025/12/22/openai-says-ai-browsers-may-always-be-vulnerable-to-prompt-injection-attacks/

https://techcrunch.com/2026/01/04/french-and-malaysian-authorities-are-investigating-grok-for-generating-sexualized-deepfakes/

https://www.bleepingcomputer.com/news/security/max-severity-ni8mare-flaw-lets-hackers-hijack-n8n-servers/

https://aws.amazon.com/security/security-bulletins/rss/2026-001-aws/

https://securityboulevard.com/2026/01/google-gemini-ai-flaw-could-lead-to-gmail-compromise-phishing-2/

https://www.scworld.com/brief/severe-ask-gordon-ai-vulnerability-addressed-by-docker

https://www.eweek.com/news/langchain-ai-vulnerability-exposes-apps-to-hack/

https://cybernews.com/security/dig-ai-new-cyber-weapon-abused-by-hackers/

https://cyberpress.org/hackers-actively-exploit-ai-deployments/