AI Briefing#ai-security
The latest AI and developer news about #ai-security, with the original source and a short summary.
The latest AI and developer news about #ai-security, with the original source and a short summary.
For voice agents to perform sensitive tasks, deterministic tool-based authentication design—rather than LLM judgment—is essential.
The US Treasury and the Fed urgently convened bank CEOs over cyber risks from Anthropic's Mythos AI.
Anthropic has announced 'Project Glasswing' together with major companies to harness AI's vulnerability-detection capabilities for defensive purposes.
It is a scanner that detects security vulnerabilities such as jailbreaks and prompt injections in GenAI models in real time.
Google is investing $12.5 million to strengthen open source security in partnership with Amazon, OpenAI, and others.
Cloudflare has released AI Security for Apps as GA and made endpoint discovery free.
As prompt injection evolves into social engineering techniques, designs are needed that go beyond simple filtering to limit the system's exposure to influence.
OpenAI is acquiring the security platform Promptfoo to strengthen the security and evaluation capabilities of AI agents.
OpenAI has released its latest case study report on detecting and preventing misuse of AI models.
Anthropic confirmed that Chinese AI companies carried out large-scale distillation attacks to illicitly extract Claude's capabilities.
Anthropic has unveiled Claude Code Security, which detects code vulnerabilities and suggests patches.
To overcome the non-deterministic limitations of AI models, Replit proposes a hybrid security model that combines static analysis with LLM reasoning.
OpenAI continuously strengthens ChatGPT Atlas's prompt injection defenses through automated red teaming techniques.
At if(kakao)25, Kakao Bank unveiled its technology in AI security, developer productivity, and data center operations.
In the Spring 2025 cycle of Amazon Research Awards, 63 recipients were selected, representing 41 universities across 8 countries.
Anthropic detected and disrupted the first large-scale cyber espionage campaign that exploited AI agent technology.
It explains the risks of prompt injection attacks that occur when AI agents reference external data, and how to respond to them.
Adopting Model Signing technology is becoming important to strengthen AI supply chain security.
Hugging Face is partnering with VirusTotal to scan all public models and datasets on the Hub.
Anthropic worked with US and UK AI security institutions to test vulnerabilities in Claude models and strengthen safeguards.
OpenAI is collaborating with US CAISI and UK AISI to examine security vulnerabilities in AI agents and strengthen its systems.
Replit has integrated automated security features directly into its platform to prevent security vulnerabilities in AI-generated code.
OpenAI is significantly expanding its cybersecurity research support and bug bounty rewards to prepare for the AGI era.
Hugging Face is adopting JFrog's scanner to enhance detection of malicious code within model weights.