The Atlantic Reports Rising Anti-AI Violence and Political Backlash
The Atlantic reports that anti-AI sentiment is escalating into real-world violence and political action. A 'Bernie-to-Bannon' coalition — Bernie Sanders and Steve Bannon — both frame AI as a disaster for the working class. In polls, the US is both the world's foremost AI developer and its chief hater.
Local data-center fights are intensifying. Maine passed the country's first statewide moratorium (vetoed). A record number of proposed data-center projects were canceled in Q1 2026 after local pushback. In extreme cases, violence has occurred:
- April: Someone shot 13 rounds at an Indianapolis councilman's home, leaving a note reading 'NO DATA CENTERS'.
- A man threw a Molotov cocktail at Sam Altman's home, then threatened to burn down OpenAI HQ. Social-media posts applauded the attack.
Political operatives are tapping into AI fears ahead of midterms. Blue Rose Research found that 'bold, populist' messaging about AI (e.g., 'fire everyone, keep profits, leave you with nothing') boosts support for Democrats. Senator Josh Hawley (R-MO) asked: 'Is it going to be good for the American worker?' Senator Mark Warner (D-VA) is 'enormously concerned' that populism could curb innovation.
Data centers, being tangible infrastructure, are easier targets than AI software for opposition. A group in Michigan published a guide titled 'How to Stop a Data Center' that advocates demonstrating outside local officials' homes. Yannick Veilleux-Lepage, a researcher, notes that AI 'generates the structural conditions historically associated with the onset of political violence.' The Soufan Center reports a rise in direct threats against AI individuals and companies, with the most common threats involving 'physical sabotage of proposed or operational data centers.'
📖 Read the full source: HN AI Agents
👀 See Also

Qwen 3.6 27B Benchmarked on DeepSWE: 2% Score, 70 Hours, 44k Avg Output Tokens
Qwen 3.6 27B (FP8, BF16 KV cache, 262k context) scored 2% on DeepSWE in 70 hours. Output tokens averaged 44k per task — comparable to larger models like Qwen 3.6 Plus. Ran on 1x RTX6000 Pro Blackwell via RunPod.

Three Inverse Laws of Robotics: Human Guidelines for AI Use
Susam Pal proposes three inverse laws of robotics for humans: don't anthropomorphize AI, don't blindly trust its output, and remain fully accountable. Practical warnings against over-reliance on generative AI.

Anthropic's Claude Mythos AI model revealed in data leak, described as 'step change' in capabilities
Anthropic is testing a new AI model called Claude Mythos (also referred to as Capybara) that represents a 'step change' in performance, with dramatically higher scores on software coding, academic reasoning, and cybersecurity tests compared to Claude Opus 4.6. The model's existence was revealed through a data leak from an unsecured, publicly-accessible data cache containing approximately 3,000 unpublished assets.

PrismML's Bonsai 1-bit Qwen models tested: 107 t/s generation on 8GB VRAM
Bonsai models from PrismML are 1-bit quantized versions of Qwen3 8B, 4B, and 1.7B that achieve 107 tokens/second generation and >1114 t/s prompt processing on an RTX 4060 with 8GB VRAM, with significantly reduced memory requirements.