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

data poisoning

topic2 events
papersTODAY 04:00 UTC

arXiv paper proposes learned selection of poison sets for LLM backdoor attacks

A new arXiv preprint introduces a method that learns which examples to poison in order to make backdoor attacks on fine-tuned language models more effective. The authors note that prior work usually holds the number of poisoned examples fixed, and their approach instead optimizes the choice of poison set. The paper appears in both cs.AI and cs.LG listings.

backdoor-attacks/llm-security/fine-tuning/research
papersTODAY 04:00 UTC

Study Reexamines How Effective Targeted Data Poisoning Attacks Really Are

A new arXiv paper argues that common evaluations of targeted data poisoning attacks are misleading because they average success rates across randomly chosen test targets, which masks worst-case outcomes. The author(s) suggest that this averaging can overstate or understate the practical threat depending on the specific samples an adversary cares about. The work calls for evaluation protocols that account for per-target variation rather than relying on aggregate scores.

data-poisoning/adversarial-ml/safety/benchmarks