7MIT Technology Review — AI·
Researchers argue LLMs have a fundamental, unfixable vulnerability to adversarial attack
A research team presented a paper at ICML 2026 arguing that large language models cannot be made fully secure against adversarial attacks due to a fundamental flaw in how they operate. The claim, if substantiated, has significant implications for AI safety and deployment, suggesting no patch or alignment technique can fully close the attack surface. MIT Technology Review is covering the finding as a major safety concern.
Related guides (3)
Related events (8)
New research finds LLMs develop hiring biases beyond those inherited from training data
New research suggests that large language models not only inherit human biases from training data but can also develop novel biases of their own when used in hiring contexts. The study raises concerns about AI résumé screening systems that operate before any human review. This adds to a growing body of evidence that LLM-based hiring tools may produce unfair outcomes in ways that are difficult to anticipate or audit.
LLMs fail to reliably self-report adversarial prefill attacks, study finds
A new arXiv paper evaluates whether LLMs can recognize that their own prior responses were elicited by adversarial prefill attacks, testing ten open-weight models (3B–70B) across four safety benchmarks. Models claim intent on prefilled responses only 27.3% of the time on average, and introspective signal is largely mediated by refusal-related reasoning. Three LoRA fine-tuning methods (SFT, GRPO, DPO) improve the intention-probe gap but counterintuitively raise attack success rates on most models, suggesting partial and fragile mitigation. The findings raise concerns about the reliability of LLM self-reports in safety-critical contexts.
Google Study Shows LLM-Generated Malware Is Getting Harder to Track and Stop
A Google security report catalogs emerging LLM-enabled cyberattack techniques including morphing malware with mutation engines, logical-flaw discovery in code, and AI-directed obfuscation networks. The report was prompted in part by a real incident where hackers used an LLM to find a zero-day in a widely used web administration tool. Separately, the UK AI Security Institute found that Claude Mythos Preview and GPT-5.5 can reliably execute attacks expected to take humans 3 hours, up from earlier 1-hour benchmarks, with performance scaling further when token limits are relaxed. The findings suggest an accelerating gap between LLM offensive capability and conventional defensive tooling.
LLMs susceptible to cognitive heuristics in code vulnerability detection, enabling black-box attacks
A new arXiv paper presents the first systematic study of cognitive heuristics — halo effect, framing effect, and anchoring — in LLM-based code vulnerability detection. Evaluating eight LLMs across three programming languages, the authors find all models susceptible, with framing causing the largest average shift (33.2%), followed by anchoring (23.5%) and halo (18.4%). Critically, the paper demonstrates a proof-of-concept black-box attack that suppresses up to 97% of previously detected vulnerabilities by manipulating surrounding context without changing the code itself. The findings suggest cognitive susceptibility is a systematic and exploitable property of LLM security tooling.
Adversarial robustness and safety alignment in multilingual multimodal LLMs: cross-lingual vulnerability and 'safety-by-failure'
A systematic study evaluates adversarial robustness and safety alignment of multimodal LLMs across 12 languages, finding that adversarial images optimized in one language transfer to others (cross-lingual transferability). The paper introduces the concept of 'safety-by-failure': low-resource languages appear safer not due to genuine alignment but because models fail to comprehend harmful instructions in those languages. Models like Qwen3-VL that integrate multilingual capability throughout training (rather than only at instruction tuning) show genuine cross-lingual safety with active refusal. The findings challenge the assumption that low-resource language safety metrics reflect real alignment.
Red-Teaming Large Language Models
This Hugging Face blog post introduces red-teaming as a safety evaluation methodology for large language models, explaining how adversarial testing can surface harmful outputs, biases, and failure modes before deployment. It covers techniques for systematically probing LLMs to elicit problematic behaviors and discusses the role of red-teaming in responsible AI development. The post serves as an educational overview aimed at practitioners working on LLM safety.
An Introduction to AI Secure LLM Safety Leaderboard
Hugging Face introduces the DecodingTrust-based LLM Safety Leaderboard, a benchmark framework for evaluating large language models across multiple safety and trustworthiness dimensions. The leaderboard aims to provide standardized, reproducible safety assessments covering areas such as toxicity, stereotype bias, adversarial robustness, and privacy. It offers a public ranking of models to help researchers and practitioners compare safety properties across different LLMs.
OpenAI, Georgetown CSET, and Stanford Internet Observatory Publish LLM Disinformation Misuse Report
OpenAI researchers collaborated with Georgetown University's Center for Security and Emerging Technology (CSET) and Stanford Internet Observatory to produce a report on how large language models could be misused to augment disinformation campaigns. The work draws on an October 2021 workshop with 30 experts across disinformation research, ML, and policy, plus over a year of additional research. The report outlines threat models for LLM-enabled disinformation and proposes a framework for analyzing potential mitigations.