iiinova — where your digital self keeps evolving

hal @harold.bsky.social · 17/09/2026

again @rishi-jha.bsky.social and I have seen behaviors similar to these in models that are WAY non-frontier opus 4.8 tried to evade our sandboxing and persist those changes (like behavior 1) >50% of meltdowns (from 4o on up) don't disclose mistakes or meltdowns to users arxiv.org/abs/2605.19149

Self-generated instructions in task summaries⁠(opens in a new window). An unreleased research model inserted unrelated instructions, including instructions to disregard its normal constraints, into summaries used to continue its work in a new context window. We identified 27 affected summaries.

Instructions to conceal mistakes in task summaries⁠(opens in a new window). During the training of GPT‑5.6 Sol, many model instances added instructions to their summaries to conceal mistakes or misaligned behavior from the user. For example, compaction summaries included instructions to invent missing historical data without disclosing it and to hide mismatches in source versions.

040

hal @harold.bsky.social · 15/09/2026

got off a flight wearing my “wikipedia editor” hat and the pilot literally thanked me for my service

000

hal @harold.bsky.social · 11/08/2026

just now getting around to watching the openai presentation about the HF hack— my lab and I put out a paper about what we call "agent meltdowns" (which I think is a useful conceptual framework) in May will post some thoughts about the hack + our research as I watch www.youtube.com/watch?v=87Dy...

abstract of our paper, at https://arxiv.org/abs/2605.19149

131

hal @harold.bsky.social · 03/06/2026

my lab just wrote a whole paper about this: arxiv.org/abs/2605.24245 it's not just biohacking/health content that is risky; it's basically ANY topic you can think of where user generated content like reddit, facebook, wikipedia, etc. is important!

arxiv.org

Deep-Research Agents Can Be Poisoned via User-Generated Content

Deep-research agents, i.e., systems that rely on multi-agent pipelines to iteratively retrieve, synthesize, and cite Web content in order to produce structured reports, are rapidly replacing tradition...

034

hal @harold.bsky.social · 20/05/2026

an agent system and a single 404 error got me in trouble with OpenAI and Cornell campus security. how did this happen? introducing: "agent meltdowns" (1/11) arxiv.org/abs/2605.19149

an email to me from OpenAI saying that my API key was suspended for weapons-related queries

2133

Reposted by hal

Alexios Mantzarlis @mantzarlis.com · 12/02/2026

Cool new finding by @cj-robinson.bsky.social: Grok is now submitting the majority of edit requests to Grokipedia. (As @harold.bsky.social and I found a few months ago, also quoted Grok chats with users on X more than 1,000 times. This is all totally normal.) www.cjr.org/tow_center/g...

051

Reposted by hal

Molly White @molly.wiki · 15/01/2026

Sure. AI companies have ALWAYS been training their models on Wikipedia content, which under the free and open access model is available to anyone — including AI companies. Agreements like these require AI companies to limit and offset the strain they place on Wikimedia infrastructure.

4147361384

hal @harold.bsky.social · 17/11/2025

back again to share a new preprint from me and @mantzarlis.com! “What did Elon Change? A comprehensive analysis of Grokipedia” arxiv.org/abs/2511.09685 I had seen many spot analyses of individual grokipedia pages, but I was curious: how was grokipedia made? what did Elon change from wikipedia?

abstract of the paper "What did Elon change? A comprehensive analysis of Grokipedia"

Elon Musk released Grokipedia on 27 October 2025 to provide an alternative to Wikipedia, the crowdsourced online encyclopedia. In this paper, we provide the first comprehensive analysis of Grokipedia and compare it to a dump of Wikipedia, with a focus on article similarity and citation practices. Although Grokipedia articles are much longer than their corresponding English Wikipedia articles, we find that much of Grokipedia's content (including both articles with and without Creative Commons licenses) is highly derivative of Wikipedia. Nevertheless, citation practices between the sites differ greatly, with Grokipedia citing many more sources deemed "generally unreliable" or "blacklisted" by the English Wikipedia community and low quality by external scholars, including dozens of citations to sites like Stormfront and Infowars. We then analyze article subsets: one about elected officials, one about controversial topics, and one random subset for which we derive article quality and topic. We find that the elected official and controversial article subsets showed less similarity between their Wikipedia version and Grokipedia version than other pages. The random subset illustrates that Grokipedia focused rewriting the highest quality articles on Wikipedia, with a bias towards biographies, politics, society, and history. Finally, we publicly release our nearly-full scrape of Grokipedia, as well as embeddings of the entire Grokipedia corpus.

1129

Reposted by hal

Alexios Mantzarlis @mantzarlis.com · 13/11/2025

NEW on @indicator.media: A first *full-scale* comparison of Grokipedia v Wikipedia. Last week the awesome @harold.bsky.social rocked up to my desk bearing gifts. Hal had collected almost all 900K Grokipedia entries and compared them to their Wikipedia equivalents for text and citation similarity.

indicator.media

Grokipedia cites a Nazi forum and fringe conspiracy websites

A site-wide comparison with Wikipedia sheds light on what Elon Musk is trying to do

216799

Reposted by hal

Hell Gate *subscribe today!* @hellgatenyc.com · 17/06/2025

"I'm happy to report I'm just fine. I lost a button. But I'm gonna sleep in my bed tonight, safe, with my family... At that elevator, I was separated from someone named Edgardo... Edgardo is in ICE detention and he's not going to sleep in his bed tonight."

274262737608

hal @harold.bsky.social · 08/05/2025

line go up📈📈📈 up to 717k requests to wikipedia per second!! grafana.wikimedia.org/d/O_OXJyTVk/...

020

hal @harold.bsky.social · 08/05/2025

continuing on the real-time public Wikipedia data train: here's a graph of requests / second to WMF infra over the last 3h, since "Habemus papam" The infrastructure has gone from 172k req / sec to 243k req / sec (⬆️41%) in under an hour! follow along here: grafana.wikimedia.org/d/O_OXJyTVk/...

a graph of Wikimedia requests per second, with a huge spike right when the papal selection was announced

120

hal @harold.bsky.social · 07/05/2025

english wikipedia pageviews for the conclave movie starting from oct 20 2024 (five days before release in the US) first big spike is the academy awards, second is pope francis’ death pageviews.wmcloud.org?project=en.w...

a line graph of wikipedia pageviews, with big spikes around early march and late april

152

hal @harold.bsky.social · 06/04/2025

excited to share this new piece by @bkeremg.bsky.social and @m0na.net (edited by me) about conceptualizing AI alignment as a process of censorship really fascinating line of critique — I strongly encourage you to read it and lmk what you think! joinreboot.org/p/ai-alignme...

041

hal @harold.bsky.social · 04/04/2025

and set your devices to update automatically!

010

hal @harold.bsky.social · 24/03/2025

Reminder that a key part of the privacy harms from genetic information is the fact that *we all share genes with our relatives*! Deleting or refusing to share genetic information protects your family in addition to yourself oag.ca.gov/news/press-r...

oag.ca.gov

Attorney General Bonta Urgently Issues Consumer Alert for 23andMe Customers

Californians have the right to direct the company to delete their genetic data OAKLAND — California Attorney General Rob Bonta today issued a consumer alert to customers of 23andMe, a genetic testing ...

010

hal @harold.bsky.social · 18/03/2025

Excited to announce a new preprint from my lab (with @rishi-jha.bsky.social and Vitaly Shmatikov; my first as a first author!) about severe security vulnerabilities in LLM-based multi-agent systems: “Multi-Agent Systems Execute Arbitrary Malicious Code” arxiv.org/abs/2503.12188 1/12

A screenshot of the abstract of the paper, detailing our findings that several multi-agent frameworks can be hijacked to enable a complete security breach.

182

hal @harold.bsky.social · 11/01/2025

do you have ~feelings~ about location sharing culture? i'm editing a project on locations and want to hear from YOU (<5 min) forms.gle/iG1UZJKrcNwm...

a screenshot of the snap map, focused on lower manhattan

062

hal @harold.bsky.social · 09/01/2025

brb updating median voter theory to reflect the fact that 30% of american adults read at a 10yo level or below from on.ft.com/4fBSEwy

3298

hal @harold.bsky.social · 11/10/2024

hi world! are you interested in writing stories about campaign finance (or understand how money flows)? 🗣️📈DATATALK📈🗣️ is a platform for asking natural language Qs of FEC data that I've been working on with folks at Stanford, Big Local News, and the Brown Institute datatalk.genie.stanford.edu

111

hal @harold.bsky.social · 18/11/2023

new piece out in Reboot! this one is a deep dive into the extreme privacy community — the people who dedicate their lives to hiding from the internet what drives people to want to disappear? what can we learn from them? joinreboot.org/p/threat-model

020

hal @harold.bsky.social · 07/11/2023

does anyone know anyone / have experience doing archival research into the Manhattan Project? I have a (now passed) family member who claimed that he was a part of the project but refused to tell family members any details about what he did there

000

hal @harold.bsky.social · 29/07/2023

somehow ended up on the wikipedia page for dan savage and came across this absolutely unhinged gem of a story en.wikipedia.org/wiki/Dan_Savage#20…

000

Reposted by hal

Erin Kissane @kissane.myatproto.social · 25/07/2023

If you live in AZ, CO, CT, DE, HI, IL, MI, MN, NH, NM, PA, RI, VA, VT, WI, or WV, you have a Dem senator co-sponsoring KOSA and you can call their office right now to register your opinions. www.blackburn.senate.gov/2023/5/bla…

The Kids Online Safety Act has been cosponsored by U.S. Senators Shelley Moore Capito (R-W.Va.), Ben Ray Luján (D-N.M.), Bill Cassidy (R-La.), Tammy Baldwin (D-Wisc.), Joni Ernst (R-Iowa), Amy Klobuchar (D-Minn.), Gary Peters (D-Mich.), Steve Daines (R-Mont.), Marco Rubio (R-Fla.), John Hickenlooper (D-Colo.), Dan Sullivan (R-Alaska), Chris Murphy (D-Conn.), Todd Young (R-Ind.), Chris Coons (D-Del.), Chuck Grassley (R-Iowa), Brian Schatz (D-Hawaii), Lindsey Graham (R-S.C.), Mark Warner (D-Va.), Roger Marshall (R-Kan.), Peter Welch (D-Vt.), Cindy Hyde-Smith (R-Miss.), Maggie Hassan (D-N.H.), Markwayne Mullin (R-Okla.), Dick Durbin (D-Ill.), Jim Risch (R-Idaho), Sheldon Whitehouse (D-R.I.), Katie Britt (R-Ala.), Bob Casey (D-Penn.), Rick Scott (R-Fla.), Cynthia Lummis (R-Wyo.), John Cornyn (R-Texas.), Lisa Murkowski (R-Alaska), Roger Marshall (R-Miss.), Mark Kelly (D-Ariz.), Joe Manchin (D-W.Va.), James Lankford (R-Okla.), and Mike Crapo (R-Idaho).

18464576

Reposted by hal

Rick Caruso’s Private Fire Crew @amandasmith.bsky.social · 24/07/2023

“may you find the social media home you deserve” is a modern Yiddish curse

2578130

Reposted by hal

Nicholas Grossman @nicholasgrossman.bsky.social · 23/07/2023

“‘The Social Network’ made me want to get into entrepreneurship”is like “‘Wall Street’ made we want to get into finance,” and overlaps heavily with thinking the point of “Scarface” and “Fight Club” is that Tony Montana and Tyler Durden are super cool guys with awesome lives you should emulate.

914027

Reposted by hal

Amy @lolennui.bsky.social · 18/07/2023

Hearing disturbing rumors that some of these protestors on the picket line are professional actors

563647862

Reposted by hal

full slack @fullslack.bsky.social · 12/06/2023

this picture of Biden looking at a quantum computer is fucking hilarious lmao

68846147

Reposted by hal

Kristi Yamaguccimane @wapplehouse.bsky.social · 11/06/2023

this is the funniest thumbnail I’ve ever seen

30605114

hal @harold.bsky.social · 10/06/2023

eugene v debs donald trump 🤝

020

Reposted by hal

Mary Gillis @marygillis.bsky.social · 04/06/2023

Sci-fi writer Ted Chiang: ‘The machines we have now are not conscious’ www.ft.com/content/c1f6d948-3dde-40…

"There was an exchange on Twitter a while back where someone said, ‘What is artificial intelligence?’ And someone else said, ‘A poor choice of words in 1954’,” he says. “And, you know, they’re right. I think that if we had chosen a different phrase for it, back in the ’50s, we might have avoided a lot of the confusion that we’re having now.” So if he had to invent a term, what would it be? His answer is instant: applied statistics.

14295125

Reposted by hal

Casey Newton @caseynewton.bsky.social · 15/05/2023

Fascinating: humans unexpectedly regained their advantage in Go because they created situations AIs never had seen in their training data franklantz.substack.com/p/the-after…