Anthropic bans 'sustained and needless abusive or cruel behavior' toward its AI models
Reported by 4 outlets from 3 newsrooms. Every outlet’s headline, how the story developed, and how loaded the writing was in each report we could read in full.
How outlets headlined it
Each outlet's own headline.
- The Register Microsoft and Anthropic play invoice tennis with startup's $17,600 Claude bill
- The Guardian Australia (Technology) Anthropic bans users from ‘needless abusive or cruel behavior’ towards Claude
- Engadget Anthropic bans 'sustained and needless abusive or cruel behavior' toward its AI models
- TechCrunch Anthropic changes usage policy to ban model abuse and election interference
How it developed
- TechCrunch filed Anthropic changes usage policy to ban model abuse and election interference
- Engadget filed Anthropic bans 'sustained and needless abusive or cruel behavior' toward its AI models
- The Guardian Australia (Technology) filed Anthropic bans users from ‘needless abusive or cruel behavior’ towards Claude
- The Register filed Microsoft and Anthropic play invoice tennis with startup's $17,600 Claude bill
Each row is the same event as one outlet told it, and the shaded row is ours: one of these reports in our plainer version, scored by the same engines, with the report it came from named. The bottom row is the spread between the outlets — our own row is excluded from it, because including it would let us widen the gap we are advertising.
Each cell shows what the outlet filed and, under it in green where it differs, what our version of that same report reads. Those are the numbers on the article page and in the extension: one article, one pair of readings, wherever you meet it.
| Outlet | Headline | Language Load | Adjectivesper 100 words | Mood |
|---|---|---|---|---|
| The Register | Microsoft and Anthropic play invoice tennis with startup's $17,600 Claude bill | 7.1 | 2.6 | 46.2 |
| The Guardian Australia (Technology) | Anthropic bans users from ‘needless abusive or cruel behavior’ towards Claude | 8.8 | 2.2 | 54.3 |
| Engadget | Anthropic bans 'sustained and needless abusive or cruel behavior' toward its AI models | 7.2 | 1.1 | 57.8 |
| TechCrunch | Anthropic changes usage policy to ban model abuse and election interference | 8.5 | 3.0 | 31.6 |
| Mundane Read our version of The Guardian Australia (Technology)’s report, which arrived with a Language Load of 8.8 | Anthropic bans users from ‘needless abusive or cruel behavior’ towards Claude | 6.1 | 1.9 | 54.3 |
| Spread | 1.7 | 1.9 | 26.2 | |
A reading is a column only when it finds something in this story, the rule that lists it beside an article. What each one means.