Bots now make up 60% of internet traffic. That number alone should set off alarm bells. AI moderation tools try to keep things tidy, but they often backfire.
On Reddit, AI removes content before humans even see it. Moderators lose context and control. One expert explained that this preemptive removal stops moderators from judging if a ban is fair. The result: communities struggle to moderate themselves effectively.
AI moderation also risks deepening societal divides. Speed and scale make mistakes worse. Instead of improving discourse, it may fuel echo chambers and mis-moderation. Avoiding social media filters out AI bots—but interacting with others can reintroduce AI contamination.
Bioengineering Meets AI: A Dangerous Mix
Large genome models now design new viruses. Not just minor tweaks—genetically distant versions of bacteria-killing viruses, called bacteriophages, are possible. This raises biosecurity alarms. Excluding viruses that infect vertebrates from training data doesn’t stop AI from replicating virus creation if it has enough resources.
Research into bacteriophages is growing, especially to fight antibiotic resistance. But AI’s ability to design novel viruses also opens doors for supervillains. They could engineer new pandemics or biological threats beyond our current defenses.
Screening DNA purchases for known pathogens no longer works. AI can create entirely new viruses that slip past these checks. This shifts the biosecurity landscape toward a dangerous unknown.
Motherboard Controllers: The Hidden Security Nightmare
Thousands of servers are vulnerable through buggy motherboard controllers called baseboard management controllers (BMCs). Over 86,000 BMCs exposed management services directly to the internet. Many come with default credentials or insecure configurations.
Exposing BMCs on the public internet or integrating them into main LANs is reckless. Open-source firmware like OpenBMC exists but doesn’t solve the problem. Security risks remain high.
Cloudflare recently open-sourced a new vibe-coding platform claiming high security. Some users remain skeptical. Meanwhile, AI tools for personal problem-solving can be secure if set up correctly. But shipping AI-generated work without accountability creates responsibility issues.
One user summed up the reckless human impulse well: “If you put a large switch in some cave somewhere, with a sign on it saying ‘End-of-the-World Switch. PLEASE DO NOT TOUCH’, the paint wouldn’t even have time to dry.”
Based on
- Coding agents can be evaluated. We just have to evaluate the work. — thenewstack.io
- AI isn’t enough to protect social media communities from AI | Ars OpenForum — arstechnica.com
- Large genome models used to design new viruses | Ars OpenForum — arstechnica.com
- Thousands of servers can be backdoored by exploiting buggy motherboard controllers | Ars OpenForum — arstechnica.com
- Cloudflare open-sources vibe-coding platform for people who aren’t coders | Ars OpenForum — arstechnica.com
