# Dr. Roman V. Yampolskiy - Monitoring Update

Accessed: 2026-08-15

## Scope

Twice-monthly maintenance check for the Managing Expectations AI section anchor:

- Page section: `ai.html#roman-yampolskiy`
- Prior source notes:
  - `research/ai/roman-yampolskiy-source-note-2026-05-18.md`
  - `research/ai/roman-yampolskiy-update-2026-08-01.md`
- Editorial rule: treat Yampolskiy's position as a serious AI-safety/risk argument, not as settled consensus or prophecy.

## Sources checked

### Official / professional sources

1. Personal/professional website
   - URL: https://www.romanyampolskiy.com/
   - Result: reachable.
   - Observed page metadata still showed `article:modified_time` / schema `dateModified` as `2026-05-19T08:05:27+00:00`.
   - No new August 2026 update marker was verified on the site during this check.

2. University of Louisville faculty page
   - URL: https://faculty.cse.louisville.edu/roman/
   - Result: reachable.
   - Visible public role remained: Roman V. Yampolskiy, Ph.D.; Associate Professor, Speed School of Engineering; Director, Cybersecurity Laboratory; Department of Computer Engineering & Computer Science.
   - No new dated update marker was verified on the faculty page during this check.

3. Google Scholar profile
   - URL: https://scholar.google.com/citations?user=0_Rq68cAAAAJ&hl=en
   - Result: reachable in this environment.
   - Visible profile metadata showed `Cited by 11,574` and the same high-citation core publications already known from earlier notes, including `The Malicious Use of Artificial Intelligence`, `Artificial Intelligence Safety and Security`, `Artificial Superintelligence`, `Artificial intelligence safety engineering`, `Responses to Catastrophic AGI Risk`, and `Leakproofing the Singularity`.
   - This check did not verify a new post-Aug. 1 paper from the visible Scholar table.

4. Official YouTube / Roman Forum feed
   - Channel: https://www.youtube.com/@RomanYampolskiy
   - Feed: https://www.youtube.com/feeds/videos.xml?channel_id=UCPIq6Bb-1iLmqyksJjy4kLQ
   - Result: active.
   - New feed entries after the Aug. 1 note were verified:
     - `Max Tegmark's P(Doom) Is Over 90%` - published 2026-08-04; URL: https://www.youtube.com/watch?v=ZO38I5UZBpg
     - `AI Safety: Pre-Training & Biosafety` - short published 2026-08-04; URL: https://www.youtube.com/shorts/Af6gK0IUczs
     - `AI's Point of No Return` - short published 2026-08-05; URL: https://www.youtube.com/shorts/7UFO6UXcJAQ
   - YouTube page/oEmbed checks verified the titles, channel, publish dates and approximate durations. The Max Tegmark episode was about 6,571 seconds and its YouTube description frames the discussion around superintelligence, AI governance, mechanistic interpretability, formal verification, recursive self-improvement and probability of losing control.

### Social / platform pages checked

- X/Twitter: https://x.com/romanyam - reachable. The public page exposed recent post metadata.
  - Aug. 4 X post: https://x.com/romanyam/status/2084660796601020870 - public text begins `After a July break, The Roman Forum is back. My new guest is Max @tegmark...` and expands to the Max Tegmark YouTube episode.
  - Aug. 11 X post: https://x.com/romanyam/status/2087285204696072212 - public text `Heretic` and expanded URL https://www.youtube.com/watch?v=tjpwAdslzHI.
- LinkedIn: https://www.linkedin.com/in/romanyam - blocked/rate-limited with HTTP 999 during this check.
- Instagram: https://www.instagram.com/romanyam/ - reachable as a platform shell; individual post extraction was not reliable from this unattended environment.
- Facebook page: https://www.facebook.com/DrRomanYampolskiy/ - reachable as a platform shell; individual post extraction was not reliable from this unattended environment.

### Search / discovery checks

- Bing web searches for recent Yampolskiy material were noisy and returned irrelevant `Roman` results, so they were not used as source evidence.
- DuckDuckGo HTML result extraction was attempted for recent Yampolskiy appearance/paper queries, but result extraction was not reliable enough to treat as evidence in this unattended run.
- Directly opened YouTube source from Yampolskiy's Aug. 11 X link:
  - `They're Summoning A DEMON - Dr. Roman Yampolskiy`
  - URL: https://www.youtube.com/watch?v=tjpwAdslzHI
  - Channel: Andrew Gold / heretics.
  - Published: 2026-08-10 according to YouTube page metadata.
  - Approximate duration: 3,670 seconds.
  - Source status: public YouTube interview/appearance; title and description use sensational framing, so quote the title only as a source title and keep Managing Expectations framing careful.

## New or newly logged public materials since 2026-08-01

### Verified official-channel material

1. `Max Tegmark's P(Doom) Is Over 90%`
   - URL: https://www.youtube.com/watch?v=ZO38I5UZBpg
   - Published: 2026-08-04.
   - Source status: official Roman Forum / Yampolskiy YouTube channel upload.
   - Relevance: meaningful new long-form Roman Forum appearance with Max Tegmark, directly aligned with Yampolskiy's AI control, superintelligence, governance, interpretability and formal-verification themes.
   - Caution: Tegmark's `P(doom)` estimate is an argument/forecast from the guest, not settled consensus or a measured probability.

2. `AI Safety: Pre-Training & Biosafety`
   - URL: https://www.youtube.com/shorts/Af6gK0IUczs
   - Published: 2026-08-04.
   - Source status: official Roman Forum / Yampolskiy YouTube short.
   - Relevance: current short-form safety/governance signal.

3. `AI's Point of No Return`
   - URL: https://www.youtube.com/shorts/7UFO6UXcJAQ
   - Published: 2026-08-05.
   - Source status: official Roman Forum / Yampolskiy YouTube short.
   - Relevance: current short-form risk framing.

### Verified public appearance material

1. `They're Summoning A DEMON - Dr. Roman Yampolskiy`
   - URL: https://www.youtube.com/watch?v=tjpwAdslzHI
   - Published: 2026-08-10.
   - Channel: Andrew Gold / heretics.
   - Source status: public YouTube interview/appearance; also linked from Yampolskiy's Aug. 11 X post.
   - Relevance: substantive new public interview after the prior Aug. 1 check.
   - Caution: the source title and description use dramatic language. For Managing Expectations, frame it as a severe AI-safety interview, not as proof that any exact catastrophic outcome is inevitable.

### Verified social/platform leads

1. Aug. 4 X post announcing the Max Tegmark Roman Forum episode
   - URL: https://x.com/romanyam/status/2084660796601020870
   - Posted: 2026-08-04T15:20:39Z according to X metadata.
   - Expanded URL: https://www.youtube.com/watch?v=ZO38I5UZBpg
   - Source status: Yampolskiy's public social post plus verified YouTube episode.

2. Aug. 11 X post sharing the Andrew Gold / Heretics interview
   - URL: https://x.com/romanyam/status/2087285204696072212
   - Posted: 2026-08-11T21:09:07Z according to X metadata.
   - Expanded URL: https://www.youtube.com/watch?v=tjpwAdslzHI
   - Source status: Yampolskiy's public social post plus verified YouTube interview page.

## Editorial update made

Meaningful new public material was found: a new official Roman Forum long-form episode with Max Tegmark, two official Roman Forum shorts, and a new Andrew Gold / Heretics interview linked by Yampolskiy. The AI page was updated with a concise latest-check paragraph and the latest source-note link now points to this Aug. 15 note.

Public framing used:

> Latest check, Aug. 15, 2026: Yampolskiy's official Roman Forum feed added a new long-form Max Tegmark episode on superintelligence risk, governance, interpretability and formal verification, plus two short clips on AI safety/biosafety and point-of-no-return framing. He also shared a new Andrew Gold / Heretics interview. Treat these as current public risk-argument materials and appearances, not proof that any exact `P(doom)` estimate or catastrophic timeline is settled consensus.

## Caution / verification labels

- `Verified`: official website reachable; University page reachable; Google Scholar reachable; official YouTube RSS reachable; Max Tegmark Roman Forum episode opened and verified via feed/page/oEmbed; Aug. 4 and Aug. 11 X metadata extracted; Andrew Gold / Heretics interview opened and verified via YouTube page/oEmbed.
- `No new official-site update`: personal website metadata still showed May 19, 2026 modification date.
- `Social lead with opened source`: X posts were used as leads, then the linked YouTube pages were opened directly.
- `Platform-limited`: Instagram and Facebook pages were reachable but not reliably extractable for individual posts in the unattended cron environment.
- `Blocked/rate-limited`: LinkedIn returned HTTP 999.
- `Framing caution`: severe AI-risk views, titles and `P(doom)` estimates should be presented as arguments/forecasts, not as settled facts or prophecy.