Fact-check a social claim before you share
Goal: Before reshare or briefing, verify social claims on a calm worksheet—screenshots are not proof.
Best for: Before reposting numbers, screenshots, or urgent-sounding news.
Social posts optimize for emotion, not evidence. Your loop restates the claim, names what would prove it, picks a primary source type, and labels Rumor when only posts exist. Models help structure the loop; they must not invent links.
Fact-check means find the primary claim, find a primary source, then judge. Screenshots alone are not sources.
Why beginners skip this
Reposting what you just scrolled past feels like a stance—without source tiers, claims drag you into misinformation. Verify before share.
Step 1 — Mini loop prompt
Before reshare, run a mini fact-check loop on one social claim; if it will not verify, delete—no “post then check.”
Claim: [paste]
1) Restate the claim in one sentence
2) What evidence would prove or disprove it
3) Best primary source type to open (official blog, regulator, company pricing, court docket, etc.)
4) If only social posts exist, label Rumor
Do not invent links. If you suggest a URL pattern, mark it as a guess to verify.
Step 2 — “Doubles prices tomorrow” screenshot
Example
Filled prompt
Claim: “Slow down viral claims with a mini verification” Write a fact-check worksheet: claim paraphrase, where first seen, primary source search plan, verdict options (true/false/mixed/unverified). Do not invent a ban.
What good output looks like
- Claim pack lists sources for “Slow down viral claims with a mini verification lo”
- Primary tab opened before share
- Unknowns labeled; no vibe citations
- Pass/fail: one primary URL saved
Step 3 — Share / wait / correct
- Share only after a primary source confirms.
- If wrong, correct in-thread; do not silently delete without a note when people already saw it.
- Pair with the 1-minute verify ritual for numbers and quotes.
Draft a 2-sentence correction if the claim was false. No sarcasm. Link the primary source I provide: [url].
Example
Filled prompt
Viral post claims Vendor X free tier includes unlimited images.
Factcheck ritual: claim → primary → support/refute/Unknown. No share until primary open.
What good output looks like
- Primary tab opened before share
- Unknowns labeled; no vibe citations
- Pass/fail: one primary URL saved
- Claim pack lists sources for “Slow down viral claims with a mini verification lo”
Save this card
Card: A screenshot is not proof; unverified claims default to no reshare. Keep this beside the tool; read before you start.
Before you start
Open a blank claim pack. One claim about “Slow down viral claims with a mini verification” plus a slot for a primary URL. Chat citations are not enough. Guide: research-factcheck-social-claims. Keep the real job visible the whole time; a blank box means stop and fetch the artifact first.
How to use this week
This week write a source tier beside three real social claims; rumors and unverified stay unshared.
Close the week by quoting aloud the refuse line you actually used once. If you cannot, rerun Practice now and date the pass/fail (Asia/Shanghai).
Keep it real
Run this guide’s Practice now on work you will actually ship this week—not a fictional scenario. Dated notes beat perfect theory.
Screenshot ≠ source
Viral images need a primary page. No page → Unverified → no share.
Do not bookmark-only
Finish Practice now, then tick this guide on your path. Titles unread are not skills.
Close the loop
Write one line: next time I repost with no source tier, I refuse.
One more pass
Re-read your Save-this-card line. If it is vague, rewrite it until a stranger could apply it.
Example
Filled prompt
Viral: "Model Y free tier now includes private SSO."
Claim → primary (pricing/changelog) → support/refute/Unknown.
No share until primary open.
What good output looks like
- Unknowns labeled; no vibe citations
- Claim pack lists sources for “Slow down viral claims with a mini verification lo”
- Pass/fail: one primary URL saved
- Primary tab opened before share
Red flag
Screenshot as proof; reshare without the mini loop—default to no reshare.
Pitfalls
| Pitfall | Fix |
|---|---|
| Asking the model “is this true?” only | Open a primary tab yourself |
| Invented citation lists | Ban invented URLs |
| Debunking with more screenshots | Prefer official text |
| Speed-reposting for engagement | Reputation > velocity |
Example
Filled prompt
Temptation: skip the habit for research-factcheck-social-claims to save five minutes.
Job still: Slow down viral claims with a mini
Smallest correct pass: Done visible · checkable artifact · Unknowns [?].
Boundaries: draft only; no silent send; no invented metrics.
Return: artifact + refuse-line said aloud.
What good output looks like
- Names the shortcut refused (specific to research-factcheck-social-claims)
- Still delivers checkable output for “Slow down viral claims with a mini”
- Unknowns labeled; draft-only
- Pass/fail: refuse-line dated (research-factcheck-social-claims #4)
Field note: run one live pass this week that proves—A screenshot is not proof; unverified claims default to no reshare. Date the pass/fail (Asia/Shanghai); bookmarks do not count.
Practice now
Screenshot provenance: save source URL with the shot (Asia/Shanghai).
Origin chase: find the earliest primary post/doc (Asia/Shanghai).
Factcheck drill: one social claim → primary source or cut (Asia/Shanghai).
Thin-spot drill: rewrite the Save card with real nouns from your work (repo, doc, inbox). Generic placeholders fail the drill.
Tonight for three real social claims write source tier and verification result; no reshare if unverified.
Buffer line: name the artifact path you used (Asia/Shanghai).
Ship-check: artifact path ___ · refuse-line dated Asia/Shanghai ___ · never-do ___
One sentence takeaway
Source tier beside every claim; rumors and unverified stay unshared.
Next
1-minute verify · Grok verify · Claim–evidence matrix · Source tiers
转发前先核实一条社交主张
目标: 转发或写进简报前,用一张冷静工作纸核实社媒主张。截图不算证明。
适合: 常在群聊或时间线看到「据说某地禁了某应用」这类断言的人。
病毒截图传得很快,一手来源却很慢。养成习惯:先复述主张,标出首次见到处,列出搜索计划,再选真 / 假 / 混杂 / 未核实。核不住,就默认不转发。
为什么新手跳过
刷到就转发像有立场;没有来源等级,主张会把你拖进假新闻。先核再转。
步骤 1——迷你循环提示词
转发前用小循环提示词拦下一条社媒主张;核不住就删,别「先发再查」。
主张:[粘贴]
1)用一句话重述主张
2)什么证据能证实或证伪
3)最好打开的一手来源类型(官网博客、监管机构、公司定价页、法院文书等)
4)如果只有社交帖,标 Rumor
不要编造链接。如果只是猜测 URL 形态,标明「待核实猜测」。
步骤 2——「明天涨价翻倍」截图
实例
填好的提示词
主张:「X 市昨天禁了应用 Y。」请写核实工作纸:主张复述、首次见到处、一手来源搜索计划、结论选项(真/假/混杂/未核实)。不要编造禁令。
边界:不编造;未知标 [?];发送前停下。
完成检查:一分钟内能判过/不过。
像样的输出大概长这样
- 把原帖主张拆成可核验的短句,每句单独标来源层级
- 找不到一级来源的主张老实标「未核实」,不硬洗成已证实
- 截图日期、账号名写进表,防止断章或张冠李戴
- 结论用「目前证据显示」措辞,不装成最终法官
- 整份核查控制在十五分钟内能交出去
步骤 3——转发 / 等待 / 更正
一手来源确认后再转发;核不住就等待或更正,别硬发。
- 如果错了,在原线程更正;已有人看到时不要默默删除不留说明。
- 数字与引语用一分钟核实仪式配合。
如果主张为假,起草两句更正。不要阴阳怪气。附上我提供的一手来源:[url]。
实例
填好的提示词
我只有一张疯传的截图,声称 X 市昨天禁了应用 Y。
请让模型列出目前还不能下的结论,以及 4 步一手来源搜索计划。禁止编造禁令。
边界:不编造;未知标 [?];发送前停下。
完成检查:一分钟内能判过/不过。
像样的输出大概长这样
- 指出哪一句把「有人发过」悄悄写成了「已被证实」
- 示范改写:补日期、补链接、加不确定标记
- 语气克制,改完的句子适合转发给同事
- 不编造你这轮搜索没找到的反证
- 收尾强化:转发前停一秒,问自己依据在哪
- 未知标出;不编造负责人、日期或指标
保存这张卡
一行可复用:记下:今天选的最薄一层 + 本月绝不加的那一层。标上日期。
开始前
聊天前用一句话写下主张;说不清主张,就是在逛,不是在核实。
本周怎么用
本周对三条真社媒主张各写来源等级;传闻与未核实一律不转发。
保持真实
用本周真要交付的工作做「现在就练一次」,别编假场景。带日期的实操笔记,比完美理论有用。
截图 ≠ 来源
病毒图片需要一手页面。没有页面 → 未核实 → 不分享。
不要只收藏
先做完「现在就练一次」,再在学习路径上打勾。只收藏不练,标题变不成技能。
闭环
写下一行:下次没来源等级就转发,我拒绝。
再过一遍
重读「保存这张卡」那一行。如果含糊,改到陌生人也能执行。
红旗
刷到就转发却没有来源等级,主张会把你拖进假新闻;先核再转。
常见坑
| 坑 | 改法 |
|---|---|
| 截图当证明 | 先填核实工作纸 |
| 核不住仍转发「先发为敬」 | 默认不发,标未核实 |
| 让模型编市政文件号 | 禁止编造;只列搜索计划 |
现在就练一次
今晚就练:对三条真社媒主张各写来源等级;传闻与未核实一律不转发。
一句话带走
来源等级旁写主张;传闻与未核实不转发。
下一步
一分钟核实 · Grok 分享前核实 · 主张–证据矩阵
← Back to home · Learn · Tools · Compare · Start