Prompt injection is an attack in which untrusted text or content attempts to override an AI application’s instructions. The malicious instruction may come directly from a user or indirectly from a webpage, document, email, or tool result processed by the model. If successful, it can cause data disclosure, unauthorized tool use, altered output, or bypassed policies. Because language models do not inherently distinguish trusted commands from quoted content, filtering suspicious phrases alone is insufficient. Defenses include isolating data from instructions, minimizing permissions, validating tool arguments, requiring approval for sensitive actions, restricting secrets, and testing with adversarial content. Applications should assume that any external content may contain instructions designed to manipulate the model.
OpenAI has launched GPT-6 Astra, a new flagship model built around computer use, coding and long-running agentic work, with major gains in mathematics, science and professional tasks.
OpenAI told lawmakers it’s developing automated shutdown capabilities for its AI systems, but a Democratic congressman says it still won’t share logs from the Hugging Face hack.
Google released Gemini 3.8 Flash for coding and reasoning at unchanged pricing, alongside a restricted cyber model, as Google engineers reportedly preferred it to Anthropic’s Opus in internal tests.
Palo Alto Networks’ CEO says AI is forcing a $1 trillion overhaul of outdated cybersecurity systems, crediting Anthropic’s Mythos model with jolting companies into urgency.
OpenAI says its unreleased Astra model can autonomously find and exploit unknown security flaws, the first system it has rated at its highest cyber-risk tier.
Anthropic has released Claude Fable 5.1, an upgraded frontier model that improves coding, research, computer use and long-running agentic work while cutting cache-read costs by 75%.
OpenAI banned ChatGPT accounts it says originated in Russia and built a fake “expert institute” using plagiarized academic work to manufacture false credibility online.
OpenAI paused reinforcement learning on its newest models after it could not rule out that an upcoming model, Astra, reached the top cybersecurity risk tier in its safety framework.
OpenAI launched ChatGPT for Teens, a version with default safety protections, learning tools and parental controls, automatically applied to users it identifies as under 18.
Anthropic’s second risk report revealed an unreleased internal model more capable than Mythos 5, and raised its catastrophic-misalignment risk rating from “very low” to “low.”