AI Flame Risk Assessment for Old Posts: Delete Only the Dangerous Ones [2026]

If you're a streamer, VTuber, or influencer, your old posts can resurface at any moment.Manually reviewing thousands of posts one by one is not realistic.

X Deleter now offers AI-powered flame risk assessment. Posts are evaluated in three levels — high, medium, and low risk — so you can target only the dangerous ones for deletion. This feature launched in June 2026.

Reviewing thousands of posts manually is nearly impossible

One VTuber had about 1,200 posts from the three years before their debut. Mixed in were harsh critiques of certain anime, rants naming former friends, and offensive jokes written late at night on impulse.

Reading through 1,200 posts individually takes days. Even then, self-review is biased — what feels like harmless banter to you may look like a serious risk to others. A post you thought was "just a joke" can become fuel for a harassment campaign.

How AI evaluates flame risk in 3 levels

The AI assessment uses Google DeepMind's Gemini model and assigns each post one of three levels with a reason for every judgment. The evaluation prompt is specifically tuned for creator-specific flame risks — sponsor criticism, aggressive replies to fans, old inappropriate jokes — weighted more heavily than general safety classification.

"Gemini is a family of multimodal large language models developed by Google DeepMind."

Source: Google AI — Gemini API https://ai.google.dev/gemini-api/docs (last confirmed: 2026-06-03)

The AI is specifically tuned for creators, streamers, and public figures. Beyond obvious red flags like hate speech, it flags creator-specific risks: sponsor criticism, aggressive replies to fans ("antis"), and old inappropriate jokes that could resurface out of context.

3 reasons AI-selective delete beats full deletion

1. Safe posts stay

Full deletion wipes everything in the date range — including fan appreciation posts, event announcements, and personal memories. AI-selective delete only removes high and medium risk posts. Everything marked "low risk" stays untouched.

2. Same price across all modes

Full delete, AI-selective, and report-only modes all cost the same: 10¢ per post (or ¥10/post in Japan), pay-as-you-go. There is no surcharge for AI analysis. The estimate is calculated from the total post count in your selected date range.

3. Full report with risk badges

Results are displayed in a clean dashboard with two tabs: "Risky (to delete)" and "Safe (kept)." Each post shows a risk badge (High/Medium/Low) and the AI's reasoning. You can verify every judgment before deletion. CSV export is available for spreadsheet analysis.

Report-only mode: assess risk without deleting

Not ready to delete? The report-only mode runs the full AI analysis and generates a complete report — but does not delete a single post. It's ideal for getting a baseline assessment of your account before deciding what to do. Same pricing applies.

"Deletes a specific Post by its ID, if owned by the authenticated user."

Source: X API — Delete Post https://docs.x.com/x-api/posts/delete-post (last confirmed: 2026-06-03)

X Deleter uses only the official X API for all operations. No unofficial scraping or browser automation. Authentication happens through X's official OAuth screen — you never enter your password on this site.

Per the X API spec, only posts owned by the authenticated user can be deleted. Other users' posts and replies to you are out of scope. Deleting your own quoted post does not remove the quote display from the other person's timeline.

Limitations: when human judgment is still needed

AI assessment is powerful but not perfect. These cases deserve your own review:

• Reply chains: Individual replies lack context. An innocent quip can be mislabeled as medium risk without the full conversation
• Inside jokes / community banter: Expressions that are fine within your community may look suspicious to the AI
• Emoji-only posts: Too little text for accurate assessment — these default to low risk

Each post in the report links to its original on X, so you can verify the AI's call with one click before committing to deletion.

How it works in 3 steps

FAQ

Does AI analysis cost extra?

No. All three modes — full delete, AI-selective, and report-only — cost the same: 10¢ per post. The estimate is based purely on post count within your selected range.

How accurate is the AI?

The AI evaluates posts with a reason for each judgment. You can review every result individually before deletion.

"Rate limits control the number of requests you can make to each endpoint."

Source: X API — Rate Limits https://docs.x.com/en/docs/x-api/rate-limits (last confirmed: 2026-06-03)

X Deleter works within this constraint, pausing after every 50 deletions and auto-resuming every 30 minutes.

Can I delete later after viewing the report?

Report-only mode does not delete anything. To delete, you'll need to run a new job with the "AI-selective delete" mode and complete payment for the same date range.

Protect your past self with AI. Start with a free estimate

No more manually sifting through thousands of posts. AI flags the fire hazards while your safe content stays untouched. AI-selective deletion keeps your safe content and removes only the fire hazards — a fundamentally different workflow from wiping everything indiscriminately.

Using only X's official API, with no password collection. Check your post count for free.