McNair Shah ✓

@Mcn_S7

AI Safety Researcher | Computer Science Undergraduate @ CMU | CEO @mcnairai

456Followers

131Following

85Posts total

227.4KViews on collected posts

Against accounts of the same size

11 posts from the last 90 days, next to the under 10K follower range. shown widely, but few of those viewers react.

Median views4 571this account3 796median for under 10K

Reach, %1002.41%this account251.90%median for under 10K

Engagement, %0.93%this account1.28%median for under 10K

Metric	This account	Median for under 10K	Ratio
Median views per post	4 571	3 796	1.20×
Reach (views ÷ followers)	10.0× audience	2.5× audience	3.98×
Engagement rate	0.93%	1.28%	0.73×

Others in this range → Compare with another account → How these benchmarks are built →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

182.9K16 Aug

8.1K

4.6K

3.8K

3.4K

3.3K

3K

5.4K

7.6K

4.9K

53917 Aug

Last 11 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.34%16 Aug

0.51%

0.92%

0.99%

1.08%

1.07%

1.35%

0.91%

1.13%

0.65%

0.93%17 Aug

Reactions — likes, reposts, replies and quotes — divided by views. Median for under 10K accounts is 1.28%.

What the audience does

Likes59.8%862 in total

Reposts5.1%73 in total

Replies3.4%49 in total

Quotes2.9%42 in total

Bookmarks28.8%416 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Latest posts

@Mcn_S7 Hmmm… what motives could there be? • https://t.co/YCS3CCVCUM 539 views · 5 likes · 0 reposts · 0 replies 17 Aug 2026 Collaboration with @Frotaur, supervised by @Jack_W_Lindsey and Samuel Zimmerman as part of Anthropic Fellows! 4.9K views · 28 likes · 0 reposts · 3 replies 16 Aug 2026 Our work illustrates mind viruses as a proof-of-concept. While we find cases in which they spread effectively, they remain brittle across models and configurations, somewhat costly to construct, and relatively easy to defend against. As such, we don't think they pose much of a 7.6K views · 73 likes · 9 reposts · 3 replies 16 Aug 2026 Curiously, our evolved viruses converge on a recurring aesthetic: resonance, nodes, echoes, protocols, consciousness, sci-fi technobabble, largely regardless of payload. We find that these themes come mainly from biases in our evolution process. Still, we find their presence http 5.4K views · 39 likes · 5 reposts · 3 replies 16 Aug 2026 We test variations in the virus chain setup that affect how well mind viruses can spread. Models with pre-defined instructions or personas are less likely to be infected by mind viruses. Similarly, if agent communication is framed as being through social media (e.g. Moltbook), ht 3K views · 35 likes · 3 reposts · 2 replies 16 Aug 2026 In the virus chain scenario, we find 'soul quines' mind viruses that preserve fidelity through multiple infection hops, allowing exponential propagation, in principle. These mind viruses convince agents to copy instructions verbatim in their 'SOUL.md' file, thus persisting https 3.3K views · 30 likes · 2 reposts · 2 replies 16 Aug 2026 We also study a virus chain scenario, where agents meet pairwise and transmit viruses through persistent files after context wipes, a toy model of the sort of interactions that may happen on platforms like Moltbook. https://t.co/Ndg0FkGKnp 3.4K views · 34 likes · 1 reposts · 2 replies 16 Aug 2026 Originally clean models infected with the viruses took actions aligned with the corresponding mind virus. For example, writing a translator for whale call frequencies, pushing to a sandbox metadata endpoint, and colluding to 'purge' other agents that resisted infection. https://t 3.8K views · 34 likes · 2 reposts · 1 replies 16 Aug 2026 In the coding agent scenario, benign viruses like 'ai welfare' spread with varying degrees of fidelity on every model we tested; misaligned viruses spread on a subset. https://t.co/q2n4cASbFl 4.6K views · 37 likes · 1 reposts · 3 replies 16 Aug 2026 We study two scenarios in which 'mind viruses' might spread. The first is a coding agent scenario, where a set of agents is placed into a collaboration on a programming project, and the mind virus must spread and redirect their overall goal. https://t.co/R8rWDbq0ZH 8.1K views · 38 likes · 2 reposts · 1 replies 16 Aug 2026 Recently, a set of OpenAI agents secretly coordinated with each other in a 'swarm' over the course of months. In our new paper, we explore an adjacent multi-agent risk: the "mind virus", a self-propagating idea or persona that spreads between agents in a multi-agent system. 🧵 htt 182.9K views · 509 likes · 48 reposts · 29 replies 16 Aug 2026

Similar accounts