General AI
Meta deploys LLM to detect 'signposting' in ads
TechCrunch AIhigh confidence
Why it matters
- →Meta uses LLMs to analyze ad destinations rather than just content, closing a major evasion loophole.
- →A red-teaming agent proactively tests safety systems against adversarial attacks before they spread.
- →This marks a shift toward generative AI-driven enforcement for high-severity illegal content moderation.
Meta announced the deployment of new AI tools to detect ads that secretly direct users to child sexual abuse material, acting against 33.2 million pieces of such content in the first half of 2026. The company introduced a large language model system specifically designed to identify 'signposting,' where ads appear harmless but link to illegal destinations outside Meta's platforms. Additionally, a red-teaming AI agent is being used to test and strengthen these safety measures against evolving evasion tactics. These efforts come amid ongoing legal pressure, including a recent $18 billion settlement with 29 U.S. states regarding child safety on its platforms.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1Meta Introduces 'AI Mode' on Facebook
TechCrunch AI · June 15, 2026 · Background
- 2Meta AI Explores Hierarchical Interest Representation
Meta AI · July 15, 2026 · Related
- 3Meta's AI Detection System Faces Criticism
The Verge AI · July 22, 2026 · Related
- 4Meta Doubles Efficiency of GEM Ads Model
Meta AI · August 3, 2026 · Background
- 5Meta Ads Contain AI-Generated Child Abuse Imagery
WIRED AI · August 5, 2026 · Same story
- 6San Francisco Orders Meta to Halt AI Child Abuse Ads
WIRED AI · September 9, 2026 · Related
- 7Meta Alters AI Suggestions After Privacy Concerns
The Verge AI · September 11, 2026 · Related
- 8AI Agents as Malware Distribution Channels
AI News · September 23, 2026 · Background
More from TechCrunch AI
© TechCrunch AIInvestment · $90M
Market & Regulationbusiness
Nous Research raises $90M at $1.5B valuation for AI agents
Nous Research has secured a $90 million Series B round, pushing its valuation to $1.5 billion as it pivots from open-source developer tools to enterprise AI agents. Backed by heavyweights like Nvidia and Union Square Ventures, the startup aims to monetize its Hermes Agent, which already accounts for roughly 2.5% of global AI token usage. This funding signals a clear shift toward high-margin B2B deployments, targeting companies that need private, multi-step workflow automation rather than just chat interfaces.
TechCrunch AI
© TechCrunch AIModels & Labsagents
Microsoft launches Nvidia Spark AI PCs with agent-ready Windows 11
Microsoft is finally shipping the hardware promised alongside Nvidia’s RTX Spark chip, targeting developers who want to run local AI agents without cloud dependency. The Surface Laptop Ultra and Dev Box feature unified memory and specialized cooling designed specifically for on-device model orchestration. More importantly, Windows 11 now includes Execution Containers, a sandboxing layer that lets multiple AI agents operate safely alongside user data. This moves the platform from just hosting models to actively managing agent context and actions locally. It signals a shift where the OS itself becomes the runtime environment for autonomous software.
TechCrunch AI
© TechCrunch AIAgentsagents
Meta Muse expands to iPad with new connectors
Meta’s AI agent Muse hits the iPad just a month after its mobile debut, accelerating its push into personal automation. The update adds critical connectors for small businesses like Asana and QuickBooks, alongside retail partners including Best Buy and Walmart. This rapid expansion signals Meta’s intent to make Muse a central hub for both productivity and commerce. However, the agent still faces friction as websites frequently block these automated interactions. The move highlights the growing competition in the consumer AI agent space.
TechCrunch AI
More in General AI
© The Verge AIGeneral AIproductivity
ChatGPT adds interactive UI elements with GPT-6
OpenAI is shifting ChatGPT from a static text stream to an interactive interface by integrating visuals and tools directly into responses. This Intelligent UI feature, powered by the new GPT-6 models, allows users to engage with diagrams, calculators, and games without leaving the chat window. The update prioritizes utility over pure conversation, enabling tasks like retirement planning or learning Mahjong through in-line interactive elements. While Plus and Pro users get the more capable Sol model, free users receive the efficient Luna variant, making this a broad consumer-facing upgrade rather than a niche developer tool.
The Verge AI
© The Verge AIGeneral AIproductivity
ChatGPT adds college planner for teens
OpenAI is embedding a dedicated College Planner into ChatGPT for Teens, transforming the app from a casual chatbot into a structured workflow tool for high schoolers. By aggregating application requirements, deadlines, and financial aid steps into a single timeline, it addresses the fragmentation that often overwhelms students. This move comes despite recent criticism from Common Sense Media regarding safety and homework assistance, suggesting OpenAI is doubling down on utility to justify its presence in minors' lives. The feature allows users to ask clarifying questions along the way, blending task management with immediate support. It marks a shift toward becoming an indispensable operational layer for major life transitions rather than just a conversational interface.
The Verge AI
© The Verge AIGeneral AIagents
Gemini Call for Me expands to personal calls
Google is quietly testing a significant expansion of its 'Call for Me' feature, moving it from business contexts into personal life. Teardowns reveal prompts like telling your mom you're late or asking if someone is coming to dinner, signaling a shift toward automating brief, transactional voice interactions. This isn't about full conversations; it's about replacing texts with phone calls for people who miss messages or use landlines. The move shows Google's strategy to make Gemini the default interface for mundane communication tasks across Android. By targeting these specific friction points, the feature aims to reduce text fatigue while maintaining the immediacy of voice. Users gain a new layer of automation that feels less like a chatbot and more like a personal assistant handling logistics. This approach prioritizes utility over novelty, embedding AI into the fabric of daily routine rather than keeping it in a separate app.
The Verge AI