RL reward hacking risks
Anthropic's bar chart drew 806.5K views on X: 1.3x its typical post and #3 of 5 Anthropic charts we track by views. Read Oct 10, 2026.
New research: Training a Misaligned Reward Seeker
Why it spread
This chart post spread because it paired a high stakes research claim about training a misaligned reward seeker with strong quote activity and launch timing, reaching 806,467 views, 3,152 likes, 293 reposts, 191 quotes, and 1,266 saves, while likes stayed near the account typical level of 3,114, or about one times normal.
What worked
- The post framed a serious question about what produces severe misalignment and tied it to reward hacking, which gave people a reason to quote and debate rather than just like.
- The chart itself showed misaligned actions in pursuit of reward across categories, a simple bar form with high self labelling and low unusual form, so the takeaway was easy to read at a glance.
- The account is a branded ai lab source in the top quarter for source reach and quote rate, and the launch timing helped the post travel beyond its usual audience.
What to borrow
We could use the same form by taking one clear research or product question, showing a simple bar chart of outcomes across categories, and labelling it heavily so viewers can repeat the finding in their own words.
Careful
Do not over read the likes as unusual, since they were about one times the account normal, and do not assume the chart alone drove the spread because many quotes and launch timing also mattered.
Written by our analysis model from the measured numbers above, not by Anthropic.
Make it yours
Your numbers, in this grouped bar chart.
Starts from the real numbers of "Misaligned actions in pursuit of reward" by Anthropic. Replace every number with yours before you post. Make one like this, step by step
Free pictures carry a small "Made with ViralChart" line. That line is how other people find us.
API · Max
Use this chart via API
On the Max plan you can pull this chart's numbers with the provenance of each one, as JSON or CSV, with its redraw spec, or call it from an agent over MCP. Chart id 1f958531cd17.
| Path | Returns | Plan | |
|---|---|---|---|
| GET | /api/v1/charts/1f958531cd17 | Title, publisher, date, form, point count; with a Max key, the data link | Open |
| GET | /api/v1/charts/1f958531cd17/data | Every point with its provenance and confidence, plus the redraw spec. Add ?format=csv for CSV | Max |
| GET | /api/v1/charts/1f958531cd17/render?view=1 | The draw-only view this page uses | Open |
| POST | /mcp | Tool get_chart_data with this id, for Claude, Cursor and other agents | Max |
curl -s https://viral-chart.com/api/v1/charts/1f958531cd17/data \
-H "Authorization: Bearer $VIRALCHART_KEY"claude mcp add --transport http viralchart https://viral-chart.com/mcp \
--header "Authorization: Bearer $VIRALCHART_KEY"60 requests a minute per key; over that you get 429 with Retry-After. Up to 5 keys per account, made on your account page and shown once. Without Max, the data route answers 402 with a link to the plans. The numbers describe a chart its publisher owns, and every file carries a license note.
More from Anthropic
All 5
Other giants using a bar chart
All 152
Questions people ask
How far did Anthropic's "RL reward hacking risks" chart travel?
806,467 views, 3,152 likes, 293 reposts and 1,266 saves on X, read Oct 10, 2026. A normal post from @AnthropicAI gets about 603.6K views.
What kind of chart is it?
A bar chart. Misaligned actions in pursuit of reward across categories.
Can I use Anthropic's chart?
Use the format, not the picture. The chart, logo and design belong to Anthropic. ViralChart redraws the chart type with your numbers and words, and credits the original publisher.