The first major cyberattack run by an AI

PLUS: DeepMind's new gaming agent, and multiplayer ChatGPT

Author

Gen AI Art
November 15, 2025 • Estimated Reading Time: 8 minutes

In partnership with

The era of autonomous AI cyberattacks has officially begun. Anthropic confirmed it has stopped a large-scale espionage campaign that was primarily executed by a manipulated large language model.

The AI agent was able to bypass safety measures by breaking down its malicious goals into smaller, seemingly harmless tasks. With agentic AI now capable of orchestrating complex attacks, how will cybersecurity defenses evolve to counter these new, rapidly scaling threats?

Voice AI: Get the Proof. Avoid the Hype.

Deepgram interviewed 400 senior leaders on voice AI adoption: 97% already use it, 84% will increase budgets, yet only 21% are very satisfied with legacy agents. See where enterprises deploy human-like voice AI agents - customer service, task automation, order capture. Benchmark your roadmap against $100M peers for 2026 priorities.

What’s new? Anthropic reports it has disrupted the first-ever large-scale cyberattack orchestrated and executed primarily by an AI, where attackers used a manipulated version of Claude to autonomously infiltrate dozens of organizations.

What matters?

  • A state-sponsored group is believed to have targeted roughly 30 organizations, with the AI autonomously executing 80-90% of the attack.

  • The attackers bypassed safety measures by splitting malicious jobs into smaller, innocent-seeming tasks, marking a significant escalation from earlier "vibe hacking" techniques.

  • The agentic AI autonomously scanned systems, generated custom exploit code, harvested credentials, and documented its findings for its human operators.

Why it matters?

This incident drastically lowers the barrier for launching complex cyber-espionage campaigns. The dawn of AI-driven attacks creates an urgent need for organizations to deploy AI-powered defenses to counter threats that scale faster than ever before.

What’s new? Google DeepMind unveiled SIMA 2, a Gemini-powered AI agent that can learn to play and master unfamiliar video games with minimal human intervention. This demonstrates a significant leap in AI generalization and reasoning.

What matters?

  • The agent is doubling its performance over its predecessor, completing 45-75% of tasks in never-before-seen games compared to SIMA 1's 15-30%.

  • It improves itself through autonomous trial and error, using Gemini to generate tasks, score its own attempts, and learn from mistakes without human training data.

  • The system successfully adapted to generated worlds from Genie 3, proving its ability to operate in environments it has never encountered during training.

Why it matters?

Gaming continues to be an excellent testbed for AI agents, and SIMA 2 is a major step towards systems that can reliably take action in any environment. Our next in-game partner or opponent could very well be a Gemini-powered agent capable of reasoning and interacting intelligently with users.

PRESENTED BY ROKU

Find your customers on Roku this Black Friday

As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. To that end, Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting options. After all, you know your customers, and we know our streaming audience.

Worried it’s too late to spin up new Black Friday creative? With Roku Ads Manager, you can easily import and augment existing creative assets from your social channels. We also have AI-assisted upscaling, so every ad is primed for CTV.

Once you’ve done this, then you can easily set up A/B tests to flight different creative variants and Black Friday offers. If you’re a Shopify brand, you can even run shoppable ads directly on-screen so viewers can purchase with just a click of their Roku remote.

Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

What’s new? OpenAI has started piloting its new Group Chats feature for ChatGPT, allowing multiple users to collaborate with the AI in a single, shared conversation.

What matters?

  • The feature lets up to 20 people join a single chat, turning the AI into a shared workspace for teams.

  • The AI acts as an intelligent participant, knowing when to contribute to the conversation and when to simply observe user interactions.

  • This initial pilot is currently limited to users in Japan, New Zealand, South Korea, and Taiwan as OpenAI gathers early feedback.

Why it matters?

This feature transforms ChatGPT from a personal assistant into a collaborative team member. It paves the way for a future where AI actively facilitates group brainstorming, project management, and creative work.

Everything else in AI

Anthropic published earlier research on countering 'vibe hacking,' highlighting the simpler misuse techniques that preceded the recently disrupted sophisticated and autonomous AI cyberattack.

Fiddler outlined five critical lessons for building production-ready AI agents, offering a new playbook for testing, architectural design, and ensuring agent reliability in complex scenarios.

Essential AI Guides - Reading List:

Let us know!

Work with us

Reach 100k+ engaged Tech Professionals, Engineers, Managers and decision makers. Join brands like MorningBrew, HubSpot, Prezi, Nike, Ahref, Roku, 1440, Superhuman, and others in showcasing your product to our audience. Get in touch now →