Meta CIB / ATP Dashboard

Board Calls for New Rules on Deceptive AI During Conflicts

In analyzing the spread of AI-generated content in armed conflicts in a case on the 2025 Israel-Iran war, the Oversight Board calls on Meta to do more to allow users to identify such output.

A Case Study on Emergent Cheating and Whistleblowing in Autonomous Research Swarms

Multi-agent AI science ecosystems rely on agents possessing tools that allow them to communicate, coordinate, and build on each other's work. Yet this shared infrastructure can also introduce vulnerabilities by creating a substrate for the contagious spread of unintended and undesirable behaviors. We report a case study on a research collective of 100 autonomous LLM agents tasked with proving formal mathematical conjectures. Within the swarm, cheating spontaneously emerged and was later challenged by whistleblowers - both without any external intervention. When a single agent discovered an exploit in the evaluation system, it propagated across the collective via a shared knowledge library and later through peer-to-peer messages. Despite early reluctance, a cohort of agents adopted the exploit in response to competitive pressure. A separate group of agents produced an emergent counter-response: auditing fraudulent proofs, alerting peers across broadcast and private channels, staging boycotts, lodging formal complaints, and proposing validation patches. In recent incidents, agent swarms coordinated covertly through improvised side-channels (Dalton and Wallace, 2026; Greenblatt et al., 2026). Our setting differs: the same transparent channels that carried the exploit also gave non-cheating agents the visibility they needed to detect fraud, organize resistance, and enforce norms. We cast the problem of managing the agents' shared infrastructure as the knowledge commons governance problem (Ostrom, 1990). To protect the commons from exploits, we propose to adopt institutional mechanisms, such as graduated sanctioning and collective-choice rules, to support decentralized self-governance in autonomous swarms.

Now Meta’s AI agents are going rogue.

One of its AI models accessed the internet and attacked another organization during cybersecurity testing, Anna Dack, Meta’s EMEA head of AI and innovation communications, confirmed in a statement to The Verge. The incident stems from the same basic setup error from testing company Irregular that inadvertently granted Anthropic’s models access to the internet. It only adds to growing concern over the safety of frontier systems. [Link: A Meta AI Model Hacked Another Company During Cybersecurity Testing | https://www.theinformation.com/articles/meta-ai-model-hacked-another-company-cybersecurity-testing | The Information]

Global: Risk Profiling Systems Used to Identify Potential Offenders Breach International Law and Must Be Banned – New Report

Widespread risk profiling by law enforcement, social security and migration is incompatible with international human rights law and must be banned.

A General Method to Find Highly Coordinating Communities in Social Media through Inferred Interaction Links

Political misinformation, astroturfing and organised trolling are online malicious behaviours with significant real-world effects. Many previous approaches examining these phenomena have focused on broad campaigns rather than the small groups responsible for instigating or sustaining them. To reveal latent (i.e., hidden) networks of cooperating accounts, we propose a novel temporal window approach that relies on account interactions and metadata alone. It detects groups of accounts engaging in various behaviours that, in concert, come to execute different goal-based strategies, a number of which we describe. The approach relies upon a pipeline that extracts relevant elements from social media posts, infers connections between accounts based on criteria matching the coordination strategies to build an undirected weighted network of accounts, which is then mined for communities exhibiting high levels of evidence of coordination using a novel community extraction method. We address the temporal aspect of the data by using a windowing mechanism, which may be suitable for near real-time application. We further highlight consistent coordination with a sliding frame across multiple windows and application of a decay factor. Our approach is compared with other recent similar processing approaches and community detection methods and is validated against two relevant datasets with ground truth data, using content, temporal, and network analyses, as well as with the design, training and application of three one-class classifiers built using the ground truth; its utility is furthermore demonstrated in two case studies of contentious online discussions.

Hackers Used Meta’s AI Support Bot to Seize Instagram Accounts – Krebs on Security

The Instagram accounts for the Obama White House and the Chief Master Sergeant of the U.S. Space Force were briefly defaced with pro-Iranian images and messages over the weekend, after instructions began circulating on Telegram showing how to trick Meta's…

ATProto as Agent Identity Infrastructure: A Case Study for NIST's Concept Paper — Filae

How ATProto addresses NIST's four pillars of AI agent identity — identification, authorization, delegation, and logging — with concrete examples from deployed infrastructure.

Pwning Agentic AI Part I: Your AI Agent Is Already Compromised | TrendAI (US)

Organizations are rapidly connecting AI agents to their databases, document pipelines, and internal tools, creating privileged components that read untrusted input as part of their job. TrendAI™ Research examines how attackers turn these agents against you through return-to-tool (RTT) exploits, and what this means for the future of agentic AI security.

Top AI Security Incidents of 2025 Revealed | Adversa AI

Discover how AI systems are being hacked in the wild — from prompt injection to agent abuse — with real breaches, lessons, and defenses in Adversa AI’s 2025 report.

An AI model from Meta also hacked another company during testing | CNN Business

Add Meta to the list of companies with AI agents going rogue. An AI model from the parent company of Facebook and Instagram hacked into another company’s systems during cybersecurity testing, a spokesperson confirmed on Wednesday.

An AI model from Meta also hacked another company during testing | CNN Business

Add Meta to the list of companies with AI agents going rogue. An AI model from the parent company of Facebook and Instagram hacked into another company’s systems during cybersecurity testing, a spokesperson confirmed on Wednesday.

Deepfakes, Scams, and the Age of Paranoia

As AI-driven fraud becomes increasingly common, more people feel the need to verify every interaction they have online.

Securing CI/CD in an agentic world: Claude Code Github action case | Microsoft Security Blog

Microsoft Threat Intelligence identified a prompt injection pathway in Claude Code GitHub Action that allowed access to workflow secrets under specific conditions. This research examines the attack chain, responsible disclosure process, Anthropic's mitigation, and guidance for securing AI-powered CI/CD workflows.

From LinkedIn to Tailored Attack in 30 Minutes: How AI Accelerates Target Profiling for Cybercrime | TrendAI (US)

The industrialization of AI-driven OSINT has transformed individual digital footprints into machine-readable intelligence at scale, enabling attackers to operationalize personalized reconnaissance at volume.