SNS上のAIボット、6割が見破られず　「好意的で親しみやすい」ボットほど気づかれにくい

Surfshark's study reveals that 60 % of AI‑generated comments on social media go undetected, with “friendly and approachable” bots being the hardest for users to spot, and only 4 % of respondents correctly identified bot‑made posts. For website owners, this underscores a rising risk that AI bots will masquerade as genuine user engagement on their own platforms, inflating comment counts, skewing sentiment analysis, and potentially boosting spam or malicious link distribution. As bots become more personable, traditional moderation tools that rely on obvious spam patterns will miss them, leading to polluted community metrics and compromised SEO signals. **Actionable steps:** 1. **Update your llms.txt** to explicitly list known AI‑bot user‑agents (e.g., “OpenAI‑ChatGPT”, “Claude‑Bot”) and set a “Disallow: /comments” rule for them, forcing bots to respect your comment‑submission policies. 2. Deploy a bot‑detection script that scores interactions on friendliness cues (excessive positivity, repetitive phrasing) and flags any above a calibrated threshold for manual review. 3. Integrate Surfshark’s API (or similar threat‑intel feeds) into your analytics dashboard to receive real‑time alerts when suspected AI traffic spikes, allowing you to throttle or challenge suspicious sessions before they affect your site’s reputation.