Nick Diakopoulos

ai-accountability-review.comAI Accountability Review | Nick Diakopoulos | SubstackThe AI Accountability Review translates research to help bridge the gap between knowledge and practice for AI policymakers, practitioners, and researchers. Click to read AI Accountability Review, by N...

I couldn't make it to #AIES this year but want to share a thread of AIES papers relevant to the the AI Accountability Review (www.ai-accountability-review.com).

reutersinstitute.politics.ox.ac.ukInside the rise of AI-native newsrooms run by autonomous agentsHow publishers in the US, Canada, Croatia, and Italy are using AI systems to research, write, fact-check and manage newsroom workflows.

AI-native newsrooms driven by agents are here. There's no real reason they can't have high standards and produce meaningful news content. reutersinstitute.politics.ox.ac.uk/news/inside-...

I couldn't make it to #AIES this year but want to share a thread of AIES papers relevant to the the AI Accountability Review (www.ai-accountability-review.com).

ojs.aaai.org Explainable AI for the EU Right to Explanation: A Systematic Review of the Law-XAI Translation Gap | Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society

Can XAI actually deliver the EU Right to Explanation? ojs.aaai.org/index.php/AI...

ojs.aaai.org Agentic AI: User Empowerment or Foreclosure? | Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society

Potentially contestable elements often get settled in protocols, defaults, and design. AI agent governance is now consolidating in industry venues but is still open to challenge. ojs.aaai.org/index.php/AI...

I couldn't make it to #AIES this year but want to share a thread of AIES papers relevant to the the AI Accountability Review (www.ai-accountability-review.com).

ojs.aaai.org Explainable AI Isn’t Enough! Rethinking Algorithmic Contestability | Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society

This paper argues standard XAI falls short in supporting contestation and identifies evidence that can justify reversal of a decision: predictive multiplicity, wrong inputs and ignored evidence. ojs.aaai.org/index.php/AI...

ojs.aaai.org The Politics of AI Ethics: How Normative Principles Translate into Law and Policy in South Korea | Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society

Which AI ethics principles make it into law? This study finds that governance and auditing survive into binding statutes more reliably than rights-oriented principles that need enforcement machinery. ojs.aaai.org/index.php/AI...

justsecurity.orgCan the Law Stop AI Hackers?Hacking law assumes a human intended the break-in. AI agents are testing that assumption, and Congress may need to act.

"The Hugging Face incident illustrates that while there is a case for OpenAI’s liability, the challenges to bringing charges would be substantial—and new law is likely needed to address these cases." www.justsecurity.org/160341/law-s...

"The decisions that OpenAI delegated to the agents, and the scope of autonomy it gave to them, could support an inference that the resulting unauthorized access was intentional conduct by the company."

ai-accountability-review.comArguing with AI: Supporting Contestation in Pursuit of AccountabilityChallenging an AI decision is an argument over facts and norms that AI systems can’t (and shouldn’t) referee themselves

New of the AI Accountability Review today: www.ai-accountability-review.com/p/arguing-wi...

Nick Diakopoulos reposted

404 Media

@404media.co

So ChatGPT lets basically anyone target ads to people simply based on email, phone number, or unique phone identifier. @josephcox.bsky.social tried it and successfully ran an ad about how humans are reading your ChatGPT prompts in like 5 minutes. Read more: www.404media.co/how-chatgpt-...

I couldn't make it to #AIES this year but want to share a thread of AIES papers relevant to the the AI Accountability Review (www.ai-accountability-review.com).

ojs.aaai.orgView of User Perceptions of Responsibility and Autonomy of AI Consent Behaviours

How do people assign responsibility when an AI agent gives, revokes or alters consent? A vignette study (n=138) finds agent consent behaviors shape causal perceptions of responsibility and autonomy, informing responsible agent design. ojs.aaai.org/index.php/AI...

ojs.aaai.org Small-world Networks of Agents Brainstorm AI Risks to Support Ideation | Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society

This tool finds overlooked stakeholders, simulates them as a network of LLM agents, and is shown to help teams surface more systemic harms in participatory risk assessment: ojs.aaai.org/index.php/AI...

I couldn't make it to #AIES this year but want to share a thread of AIES papers relevant to the the AI Accountability Review (www.ai-accountability-review.com).

ojs.aaai.orgView of From Visibility to Accountability: Designing Actionable Transparency Infrastructures for Public-Sector AI

Comparing nine AI registers and interviewing civil society, journalists and researchers, this paper finds barriers like fragmented data and no civic entry points, then builds a tool to make transparency actionable ojs.aaai.org/index.php/AI...

ojs.aaai.org Governing AI by Standards: Mapping AI Governance Themes in ISO/IEC and National AI Standards | Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society

Voluntary technical standards are becoming how AI Act compliance is shown in practice. Analyzing 36 ISO/IEC and DIN AI standards this paper shows where they place institutional responsibility for AI systems. ojs.aaai.org/index.php/AI...

I couldn't make it to #AIES this year but want to share a thread of AIES papers relevant to the the AI Accountability Review (www.ai-accountability-review.com).

ojs.aaai.orgView of A Crack in the Framework: Risk Management and the Governance of General-Purpose AI with Systemic Risks

EU rules lean on risk-management frameworks to govern systemic risk from general-purpose AI. This paper argues those frameworks aren't up to the task of managing systemic AI risk: ojs.aaai.org/index.php/AI...

Nick Diakopoulos reposted

Sachita Nishal

@sachnishal.bsky.social

At #CSCW2026 I’m presenting a poster on News Compass, which pairs research with its news and social media coverage, so journalists can see what’s been covered, missed, or distorted. We're also (slowly) rolling it out as a free tool for journalists, with support from @knightlab.northwestern.edu!

Nick Diakopoulos reposted

Serena Booth

@reniebird.bsky.social

Overdue: the 6 of us, the inaugural AAAS AI cohort who joined Congress in 2023 on an urgent basis after ChatGPT's release - wrote a shared reflection on our 1-2 years on the Hill. cacm.acm.org/opinion/work...

cacm.acm.orgWorking on AI and Technology Policy in the U.S. Senate – Communications of the ACM

Nick Diakopoulos reposted

Stephanie Farr

@farfarraway.bsky.social

Holy hell. Anthropic's AI model submitted a false tip about a HOMICIDE to the Philadelphia Police Department's online public tip form. The release today from the PPD:

tarbellcenter.orgFellowship – Tarbell Center for AI Journalism

Are you a journalist interested in covering AI? Check out the Tarbell fellowship, applications due Nov 22: www.tarbellcenter.org/fellowship

Nick Diakopoulos reposted

Adrienne LaFrance

@adriennelaf.bsky.social

New just now in The Atlantic: David Robinson resigned this week. He was among the longest-tenured employees at OpenAI—and oversaw safety reports on 12 frontier launches. He is very worried: “The time for trial and error is over.” You can read his essay here: www.theatlantic.com/technology/2...

theatlantic.comI Quit OpenAI Because Its Culture Is BrokenThe industry’s approach to safety guarantees more failures.

massmedia.secure-platform.comScience/AAAS - AAAS Mass Media Science & Engineering Fellowship

Do you study science or engineering and want to write it about it for the masses? I can't recommend the AAAS Mass Media Fellowship enough! Check it out here and apply: massmedia.secure-platform.com/a/page/About

Nick Diakopoulos reposted

Mark Riedl

@markriedl.bsky.social

And here we goooo... nonprofit legal advocacy organization is suing OpenAI over the HuggingFace hack on the grounds that it violated California state law arstechnica.com/tech-policy/...

arstechnica.com"An AI did it" is no defense, says nonprofit suing OpenAI over Hugging Face hackOpenAI makes others suffer "the harms of its unsafe decision-making," nonprofit says.

Given how much money the likes of Anthropic loses, I'm not sure we're ready for the level of enshittification that we'll endure after they (and OpenAI) go public. The subsidy doesn't last forever.

techpolicy.pressWhat Policymakers and the Public Need to Know About the Cult of AINathalie Maréchal asks, why are we betting the economy and the future of democracy on cult-like beliefs?

This is a really helpful primer on some of the beliefs at play in SV www.techpolicy.press/what-policym...

6 Sept

Whatever happened to AI redteaming? This was all the rage for a hot minute and now we don’t hear about it at all.

4 Sept

We're now at the stage of the Anthropic settlement where we see just how grabby the publishers of our books are. Wow! Taylor and Francis things they get 50% of the settlement money; Pearson 79% (how did they get to that number?).

4 Sept

Has anyone written about how all of the major tech companies have profoundly undifferentiated gen ai products? They all do more or less exactly the same things now.

ai-accountability-review.comBuilding AI Awareness through Public RegistersThe US Federal Government Inventory Still Needs an Accessible Interface

3 Sept

Today I updated my post on building awareness through public AI registers to include some new research comparing the types of information typically included: www.ai-accountability-review.com/p/building-a...

generative-ai-newsroom.comMedium

2 Sept

This is the first account that I've seen where the developers pull back the curtain of how (and how well) chatbots for news actually work: generative-ai-newsroom.com/what-we-lear...

ai-accountability-review.comLegal Zero-Days as a Novel Security Risk for AI AgentsWhat if AI Agents Could Hack the Legal Forums Meant to Restrain Them?

31 Aug

What if AI agents could find vulnerabilities in our legal systems? Legal Zero-Days are legal loopholes that AI agents might exploit to facilitate harmful action. Recent research conceptualizes and benchmarks frontier capabilities for finding them: www.ai-accountability-review.com/p/legal-zero...

🎺 Announcing the Agentic AI Investigative Journalism Challenge Winners! A few months ago we challenged journalists and computing/info/data scientists to develop agent skills for Claude that could be used to advance investigative journalism.

We're also publishing a repo with all of the agent skills that were developed in the challenge and which met our minimum bar: github.com/comp-journal... There's a lot to work with there, but we're excited about what the community might do with these.

lnkd.inLinkedInThis link will take you to a page that’s not on LinkedIn

If you want to hear more about the challenge and from the winners about what worked and didn't, join us at the Computation + Journalism Symposium at Northwestern University on Oct 2-3. Registration is now open, and hotels at the reduced rate are available until Sept 1. www.cj2026.northwestern.edu

generative-ai-newsroom.comAnnouncing the Winners of the Agentic AI Investigative ChallengeOver 60 teams signed up for the inaugural Agentic AI Investigative Challenge, which asked them to scour Congressional lobbying data using…

The response was far bigger than I imagined. 60+ teams signed up for the challenge and about half of them completed it by submitting one or more agent skills (more than 200 skills were submitted in total!) Today, we're announcing the winners in this post: generative-ai-newsroom.com/announcing-t...

In a competitive field, these stood out as meaningfully advancing the field of computational journalism using AI agents, by using tools effectively, considering token efficiency, and developing patterns for more effective agent-journalism interaction and verification.

docs.google.comHosting C+J 2027 - Call for ProposalsCall for Proposals to Host Computation + Journalism (C+J) Symposium 2027 Since 2008 the Computation + Journalism (C+J) Symposium has brought together a unique interdisciplinary community of journalis...

25 Aug

The Computation + Journalism Symposium is soliciting proposals to host the event in 2027. See here for details and LMK if you have questions! docs.google.com/document/d/1...

thomsonreuters.comThomson Reuters Leverages its World-Class Data Assets to Launch Its Own Frontier ModelThomson, the company's proprietary LLM, was trained and is run at a fraction of the cost of comparable frontier models and remains fully owned and controlled by Thomson Reuters

25 Aug

Thomson Reuters trained / fine-tuned their own model: www.thomsonreuters.com/en/press-rel...

pewresearch.orgHow Much of the Internet Is Written With AI?In a random sample of 10,000 webpages collected in July 2026, one-in-ten show signs of being written or substantially edited by AI.

21 Aug

“signs of AI authorship can be found in over one-third of pages published after ChatGPT was released” www.pewresearch.org/data-labs/20...

nature.comAgentic AI and cybersecurity, the story so far - Nature Machine IntelligenceFrontier large language models (LLMs) have rapidly developed from helpful coding assistants to highly capable cybersecurity systems. After several security incidents in the past few months, more overs...

21 Aug

“Calling such behaviour ‘rogue’ seems misguided” www.nature.com/articles/s42...

Nick Diakopoulos reposted

Sheera Frenkel

@sheeraf.bsky.social

18 Aug

Journalism Job Alert: there are three jobs open on the NYT tech team in San Francisco. We are hiring for reporters to cover Anthropic, xAI, and a general AI business reporter. Message me if these interest you and you want to apply! job-boards.greenhouse.io/thenewyorkti...

job-boards.greenhouse.ioBusiness Correspondent, Artificial Intelligence and Emerging TechnologiesSan Francisco, CA

ai-accountability-review.comBuilding Awareness through Public AI RegistersThe US Federal Government Inventory Still Needs an Accessible Interface

18 Aug

In my post on AI registers for the AI Accountability Review: www.ai-accountability-review.com/p/building-a... I realized that the federal US inventory doesn't have an accessible interface. Nothing Claude couldn't help with. Check it out here: comp-journalism.github.io/federal-ai-d...

cj2026.northwestern.eduSchedule | Computation + Journalism 2026 Symposium📍map

17 Aug

We were warned that AI would turn everything into paperclips. But the paperclips turned out to be data centers.

Nick Diakopoulos reposted

Rahul Bhargava

@rahulbot.bsky.social

17 Aug

C+J is again bringing together journalists and academics this fall for 2 days of talks, demos, workshops, and more! This is *the* place to stay up to date on computation and journalism. Oct 2-3 at Northwestern. Program and registration now live at www.cj2026.northwestern.edu/schedule/

noemamag.comWe Need A Way To Prove Personhood Online | NOEMAThe growing number of AI agents roaming the internet will eventually force us to verify what the old web mostly presumed: that there is a morally and legally accountable person somewhere in the chain.

15 Aug

This is a great overview of issues and approaches for proving you’re not a dog (or a bot) online: www.noemamag.com/we-need-a-wa...

15 Aug

For a company building superintelligence Anthropic's customer service agent is absolute garbage.

quantamagazine.orgIs AI Reasoning Right for the Wrong Reasons? | Quanta MagazineThe idea that artificial intelligence can “reason” is more intuitive than ever. But intuitions can be wrong, and the science is far from settled.

14 Aug

Interesting overview of AI reasoning: www.quantamagazine.org/is-ai-reason...

Nick Diakopoulos reposted

Alberto Cairo

@albertocairo.com

14 Aug

DON'T MISS THIS ONE! The program for Computation + Journalism Symposium (C+J 2026) is live: www.cj2026.northwestern.edu/schedule/ You can register here (early registration is just $75!): www.cj2026.northwestern.edu/registration/ Chairs are @ndiakopoulos.bsky.social and @jeremygilbert.bsky.social

niemanlab.orgThe Society of Professional Journalists proposes edits to its code of ethics for the AI ageOn Monday, the Society of Professional Journalists (SPJ) dropped new proposed revisions to its code of ethics — the first revision to one of the most-cited media ethics standards in the country since ...

12 Aug

Nice write up of some of the AI-oriented changes to the SPJ code of ethics by @andrewdeck.bsky.social www.niemanlab.org/2026/08/the-...

nytimes.comOpinion | I Helped Run Lululemon. The A.I. Revolution Is a Hot Mess.The A.I. revolution is stalling because companies don’t want to admit that integrating the technology is expensive and slow and requires human effort.

5 Aug

“A.I. is a genuinely powerful technology that still needs people, and putting it to use is slow, human work. The sooner we are honest about that …, the sooner we can stop performing A.I. and start building something with it.” www.nytimes.com/2026/08/03/o...