AI companion privacy and safety — Last 30 Days Roundup

• Date range: 2026-10-03 to 2026-11-02
• Sources: 2 active (GitHub, Hacker News)

Warnings

• Top evidence is highly concentrated in one source.
• Some sources failed: grounding, reddit, youtube

Ranked Evidence Clusters

1. AI Safety Is Mostly a Sex Cult in Berkeley, California (score 47, 1 item, sources: Hacker News)

• [hackernews] AI Safety Is Mostly a Sex Cult in Berkeley, California

- 2026-10-23 | Hacker News | [29pts, 5cmt] | score:47 - URL: https://www.verysane.ai/p/ai-safety-is-mostly-a-sex-cult-in-berkeley-california - Evidence: AI Safety Is Mostly a Sex Cult in Berkeley, California

2. AI Safety Is Not a Corporate Plot (score 45, 1 item, sources: Hacker News)

• [hackernews] AI Safety Is Not a Corporate Plot

- 2026-10-15 | Hacker News | [18pts, 22cmt] | score:45 - URL: https://www.astralcodexten.com/p/ai-safety-is-not-a-corporate-plot - Evidence: AI Safety Is Not a Corporate Plot

3. Show HN: Spring AI Privacy Guardrails – PII, tool permissions, content checks (score 45, 1 item, sources: Hacker News)

• [hackernews] Show HN: Spring AI Privacy Guardrails – PII, tool permissions, content checks

- 2026-10-27 | Hacker News | [3pts, 1cmt] | score:45 - URL: https://github.com/ultramancode/spring-ai-privacy-guardrails - Evidence: Show HN: Spring AI Privacy Guardrails – PII, tool permissions, content checks

4. Anthropic proposes AI safety bill with $1B / 50-death treshold (score 45, 1 item, sources: Hacker News)

• [hackernews] Anthropic proposes AI safety bill with $1B / 50-death treshold

- 2026-10-18 | Hacker News | [6pts, 1cmt] | score:45 - URL: https://www.politico.com/news/2026/10/16/anthropic-ai-safety-01124654 - Evidence: Anthropic proposes AI safety bill with $1B / 50-death treshold

5. The focus on AI safety disasters ignores everyday harms (score 44, 1 item, sources: Hacker News)

• [hackernews] The focus on AI safety disasters ignores everyday harms

- 2026-10-27 | Hacker News | [3pts] | score:44 - URL: https://www.technologyreview.com/2026/10/26/1147645/the-focus-on-ai-safety-disasters-ignores-everyday-harms/ - Evidence: The focus on AI safety disasters ignores everyday harms

6. Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages (score 44, 1 item, sources: Hacker News)

• [hackernews] Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages

- 2026-10-18 | Hacker News | [172pts, 167cmt] | score:44 - URL: https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropics-proposed-ai-safety-bill-only-kicks-in-if-50-deaths-or-usd1-billion-in-damages-are-on-the-line-proposal-would-effectively-make-the-commerce-department-the-only-agency-capable-of-stopping-an-ai-model - Evidence: Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages

7. Regulate the deployment, not just the model: a who-does-what plan for AI safety (score 42, 1 item, sources: Hacker News)

• [hackernews] Regulate the deployment, not just the model: a who-does-what plan for AI safety

- 2026-10-24 | Hacker News | [3pts] | score:42 - URL: https://substack.norabble.com/p/ai-safety-is-all-hands-on-deck - Evidence: Regulate the deployment, not just the model: a who-does-what plan for AI safety

8. OpenAI fires 3 safety researchers in dispute over AI risks (score 41, 1 item, sources: Hacker News)

• [hackernews] OpenAI fires 3 safety researchers in dispute over AI risks

- 2026-10-11 | Hacker News | [15pts] | score:41 - URL: https://apnews.com/article/openai-chatgpt-ai-artificial-intelligence-safety-789d4f5293fba45a22fcb62ebfbc2a41 - Evidence: OpenAI fires 3 safety researchers in dispute over AI risks

9. Workspace AI data policy (EU-only, zero data retention) and Privacy settings tab (score 40, 1 item, sources: GitHub)

• [github] Workspace AI data policy (EU-only, zero data retention) and Privacy settings tab

- 2026-10-19 | twentyhq/twenty | [2react, 8cmt] | score:40 - URL: https://github.com/twentyhq/twenty/pull/28296 - Evidence: \#\# What

Workspaces can restrict AI to models whose route states an EU residency and/or zero data retention (ZDR). The setting lives in a new Settings > General > Privacy tab, which also presents how Twenty Cloud protects data.

A route that does not state its residency or retention never qualif

10. AI's quiet safety gatekeepers are stepping into the spotlight (score 39, 1 item, sources: Hacker News)

• [hackernews] AI's quiet safety gatekeepers are stepping into the spotlight

- 2026-10-13 | Hacker News | [5pts] | score:39 - URL: https://www.cnbc.com/2026/10/11/ais-quiet-safety-gatekeepers-are-stepping-into-the-spotlight.html - Evidence: AI's quiet safety gatekeepers are stepping into the spotlight

11. AI safety money floods into politics (score 39, 1 item, sources: Hacker News)

• [hackernews] AI safety money floods into politics

- 2026-10-14 | Hacker News | [4pts] | score:39 - URL: https://www.axios.com/2026/10/14/ai-safety-regulation-election-spending-50-million - Evidence: AI safety money floods into politics

All Items by Source

Hacker News (11 items)

50141429 (score:0) f-serif (2026-10-18) [172 points, 167 comments] Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropics-proposed-ai-safety-bill-only-kicks-in-if-50-deaths-or-usd1-billion-in-damages-are-on-the-line-proposal-would-effectively-make-the-commerce-department-the-only-agency-capable-of-stopping-an-ai-model Hacker News Anthropic's proposed AI safety bill only kicks in if 50 deaths or $1B in damages

50211618 (score:0) SLHamlet (2026-10-23) [29 points, 5 comments] AI Safety Is Mostly a Sex Cult in Berkeley, California https://www.verysane.ai/p/ai-safety-is-mostly-a-sex-cult-in-berkeley-california Hacker News AI Safety Is Mostly a Sex Cult in Berkeley, California

50106521 (score:0) jefftk (2026-10-15) [18 points, 22 comments] AI Safety Is Not a Corporate Plot https://www.astralcodexten.com/p/ai-safety-is-not-a-corporate-plot Hacker News AI Safety Is Not a Corporate Plot

50255993 (score:0) joozio (2026-10-27) [3 points] The focus on AI safety disasters ignores everyday harms https://www.technologyreview.com/2026/10/26/1147645/the-focus-on-ai-safety-disasters-ignores-everyday-harms/ Hacker News The focus on AI safety disasters ignores everyday harms

50259632 (score:0) ultramancode (2026-10-27) [3 points, 1 comments] Show HN: Spring AI Privacy Guardrails – PII, tool permissions, content checks https://github.com/ultramancode/spring-ai-privacy-guardrails Hacker News Show HN: Spring AI Privacy Guardrails – PII, tool permissions, content checks

50220831 (score:0) nedruod (2026-10-24) [3 points] Regulate the deployment, not just the model: a who-does-what plan for AI safety https://substack.norabble.com/p/ai-safety-is-all-hands-on-deck Hacker News Regulate the deployment, not just the model: a who-does-what plan for AI safety

50263325 (score:0) JumpCrisscross (2026-10-27) [39 points, 10 comments] Police Are Pulling People over Because AI Deems Their Driving "Suspicious" https://www.aclu.org/news/privacy-technology/ai-driven-police-stops Hacker News Police Are Pulling People over Because AI Deems Their Driving "Suspicious"

50146270 (score:0) satirr (2026-10-18) [6 points, 1 comments] Anthropic proposes AI safety bill with $1B / 50-death treshold https://www.politico.com/news/2026/10/16/anthropic-ai-safety-01124654 Hacker News Anthropic proposes AI safety bill with $1B / 50-death treshold

50093725 (score:0) worik (2026-10-14) [4 points] AI safety money floods into politics https://www.axios.com/2026/10/14/ai-safety-regulation-election-spending-50-million Hacker News AI safety money floods into politics

50080408 (score:0) gmays (2026-10-13) [5 points] AI's quiet safety gatekeepers are stepping into the spotlight https://www.cnbc.com/2026/10/11/ais-quiet-safety-gatekeepers-are-stepping-into-the-spotlight.html Hacker News AI's quiet safety gatekeepers are stepping into the spotlight

50047557 (score:0) devonnull (2026-10-11) [15 points] OpenAI fires 3 safety researchers in dispute over AI risks https://apnews.com/article/openai-chatgpt-ai-artificial-intelligence-safety-789d4f5293fba45a22fcb62ebfbc2a41 Hacker News OpenAI fires 3 safety researchers in dispute over AI risks

GitHub (2 items)

GH10 (score:0) FelixMalfait (2026-10-19) [8 comments] Workspace AI data policy (EU-only, zero data retention) and Privacy settings tab https://github.com/twentyhq/twenty/pull/28296 twentyhq/twenty \#\# What

Workspaces can restrict AI to models whose route states an EU residency and/or zero data retention (ZDR). The setting lives in a new Settings > General > Privacy tab, which also presents how Twenty Cloud protects data.

A route that does not state its residency or retention never qualif

GH7 (score:0) shaiss (2026-10-12) [32 comments] Coach-owned adaptive targets: AI sets and adjusts daily goals (#70) https://github.com/shaiss/StreamDeckHabitTracker/pull/93 shaiss/StreamDeckHabitTracker \#\# Summary

Coach-owned adaptive targets (#70): the AI proposes/adjusts daily goals (downs auto-apply; ups queue for human approve/dismiss), with locked writes across habits:config + habits:targets.

\#\# CodeRabbit status (0f6907c)

Re-reviewed **2026-10-14 ~

Stats

• Total evidence: 13 items across 2 sources
• Top voices: Hacker News, twentyhq/twenty, shaiss/StreamDeckHabitTracker
• GitHub: 2 items | 5react, 40cmt | voices: twentyhq/twenty, shaiss/StreamDeckHabitTracker
• Hacker News: 11 items | 297pts, 206cmt | domains: Hacker News

Source Coverage

• GitHub: 2 items
• Web: 0 items (unreachable: Keyless web search unavailable (run doctor for fixes))
• Hacker News: 11 items
• Reddit: 0 items (timeout: URL Error: timed out (run doctor for fixes))
• YouTube: 0 items (timeout: Search timed out after 120s (run doctor for fixes))

Source Errors

• Web: Keyless web search unavailable
• Reddit: URL Error: timed out
• YouTube: Search timed out after 120s

✅ All agents reported back! ├─ 🟡 HN: 11 storys │ 297 points │ 206 comments ├─ 🐙 GitHub: 2 items │ 5 reactions │ 40 comments ├─ 🕒 Recent evidence is thin: only 3 of 13 dated items are from the last 7 days. └─ 📎 Raw results saved to ~/.config/last30days/ai-companion-privacy-and-safety-raw-2026-11-02.md