X becoming a 'ghost town' of bots as AI-generated spam content floods internet
Users are split over whether X (formerly Twitter) is being overrun by bots and AI‑generated spam or remains usable, with experiences differing sharply between small, niche circles and replies to viral or high‑profile posts. Many blame the platform’s pay‑for‑engagement model, blue‑check promotion, and staff cuts for incentivizing spam and weakening moderation, while others report fewer crypto scams and better conversations than in past years. Broader concerns emerge about how any user‑generated platform can resist increasingly sophisticated AI content and bot networks without heavy‑handed identity verification or costly human moderation.
Perceived Scale of the Bot/Spam Problem
- Many report replies to popular tweets being “overrun” by bots, AI‑generated spam, crypto scams, OnlyFans/NSFW spam, and engagement-farming replies unrelated to the original post.
- Others say they see very few or no bots, especially when following smaller accounts and avoiding viral threads.
- Some claim spam (especially crypto) is lower than pre‑Musk, others say bot and scam activity is much worse now.
- Experiences differ sharply by “side of Twitter” (tech vs sports/entertainment vs politics, small vs big accounts).
Suspected Causes and Incentives
- Frequent claim: layoffs of anti‑spam staff and shutdown of automated detection tools degraded defenses.
- Others emphasize monetization/product changes: pay‑for‑attention, revenue sharing, and blue checks boosted incentives for spam and made bot replies more visible.
- Self‑hosted LLMs and cheap automation are seen as enabling large-scale content and engagement bots.
- Some suggest management may tolerate bots because they create the illusion of activity and help ad metrics.
User Experience and Community Shifts
- Some find X “hollowed out”: more bot followers, fewer genuine interactions, low reach for quality posts, and feeds dominated by scams, low-quality ads, or “For You” junk.
- Others report the opposite: better conversations, faster spam removal, and an overall improved experience vs 2016–2021, especially around politics and “cancel culture.”
- There’s disagreement about whether pre‑Musk moderation suppressed certain views; examples are debated and contested.
Mitigation Strategies and Platform Design
- Proposed personal strategies: avoid new or lopsided accounts, aggressively mute/block suspicious profiles, stay in smaller/focused communities, ignore popular reply chains.
- Concerns that serious bot crackdowns would require rehiring staff and could expose the extent of past bot traffic, hurting metrics and advertiser confidence.
- Some hope identity verification/blue checks will eventually separate real from fake; others note paid blue bots are already a major problem.
Broader Implications
- Several argue this is an internet‑wide issue: any UGC platform will face endless LLM‑driven spam.
- Hard-ID or state-issued ID systems are discussed as a possible but unappealing solution.
- HN is cited as a contrast: low incentives, active moderation, and community downvoting keep bot/LLM content in check—for now.