LIVE PULSE
4.0 Anthropic CEO Amodei calls for slower AI development and shared safety rules — 11 src2.2 Agility Robotics unveils Digit 5 humanoid for warehouses and factories — 2 src2.0 Apple ships rebuilt Siri with Google Gemini, but not in the EU — 2 src1.8 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions — 2 src1.4 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns — 5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says — 2 src1.1 Anthropic data retention policy prompts firms to limit Claude use for sensitive work — 1 src1.1 Study examines issue bias in LLMs used as writing assistants before Swedish 2026 election — 1 src1.1 Study Audits Misalignment in Multi-Modal World Models — 1 src1.1 Retrieval-Grounded Reasoning Approach Proposed for Universal Multimodal Embeddings — 1 src4.0 Anthropic CEO Amodei calls for slower AI development and shared safety rules — 11 src2.2 Agility Robotics unveils Digit 5 humanoid for warehouses and factories — 2 src2.0 Apple ships rebuilt Siri with Google Gemini, but not in the EU — 2 src1.8 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions — 2 src1.4 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns — 5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says — 2 src1.1 Anthropic data retention policy prompts firms to limit Claude use for sensitive work — 1 src1.1 Study examines issue bias in LLMs used as writing assistants before Swedish 2026 election — 1 src1.1 Study Audits Misalignment in Multi-Modal World Models — 1 src1.1 Retrieval-Grounded Reasoning Approach Proposed for Universal Multimodal Embeddings — 1 src
HEATPULSEAI MAGAZINES
FLIP · FOLLOW · SAVE
Log inSign up

backdoor-attacks

topic2 events
papersTODAY 04:00 UTC

Backdoor attacks found against pretrained latent world models used for control

A new arXiv paper examines how pretrained world models, which learn latent representations of observations and predict their evolution under actions, can be compromised by backdoor attacks when reused as general-purpose dynamics backbones for control tasks. The authors study the security risks this reuse creates for downstream control systems.

safety/world-models/backdoor-attacks/research
papersTODAY 04:00 UTC

Quantization-Conditioned Backdoor Attacks Target Open-Weight LLM Agents

A new arXiv paper describes an attack in which an adversary releases a full-precision model checkpoint that passes standard audits but behaves maliciously once it is quantized for deployment. Because quantization is a common default path for running open-weight agent models, the technique could let compromised agents slip past pre-release checks. The work frames this as a supply-chain risk for quantized LLM deployments.

safety/agents/quantization/backdoor-attacks