iiinova — where your digital self keeps evolving

Reposted by Andrew Porter

Lisa Messeri @lmesseri.bsky.social · 09/10/2026

“Perhaps the most relevant issue is that by releasing this trove of material in this manner, the companies are in the process of destroying the ecosystem that made it possible for this work,” she said." To me, this is the crux. Science is a human, community endeavor. They are destroying science.

nytimes.com

‘Breathtaking,’ ‘Devastating’: Mathematics Reels After New OpenAI Release (Gift Article)

Hundreds of new A.I.-generated findings moved the frontiers of higher math in a single day — dispelling any doubt that the field is forever changed.

48841

Andrew Porter @retropz.bsky.social · 06/10/2026

"The assumption was that if you buy an antibody against a particular protein, it will recognise that target...if you buy a bag of sugar from the supermarket for example, you wouldn’t expect it to turn out to be salt." References work from @reeserichardson.bsky.social & @sholtodavid.bsky.social

062

Reposted by Andrew Porter

Loish @loish.bsky.social · 03/10/2026

Dang yall, thanks so much for the support 😭 Usually I ignore the random AI accusations but I wanted to take the chance to talk about how important it is to do your research in times like these. Didn't expect to get soooo many kind comments though! We gotta stick together, my beloved art community ♥️

452717470

Reposted by Andrew Porter

Andrew Gelman et al. @statmodeling.bsky.social · 02/10/2026

If you were planning to send me a chatbot-written email or post a chatbot-written comment, just send or post your prompt. That’s enough. You can keep the slop to yourself. statmodeling.stat.columbia.edu/2026/10/02/i...

statmodeling.stat.columbia.edu

If you were planning to send me a chatbot-written email or post a chatbot-written comment, just send or post your prompt. That’s enough. You can keep the slop to yourself. | Statistical Modeling, C...

0253

Reposted by Andrew Porter

Randall Munroe @xkcd.com · 01/10/2026

If I went to Switzerland and won a Swiss watch precision contest, I would literally never talk about anything else

568413843

Reposted by Andrew Porter

maura quint @mauraquint.bsky.social · 01/10/2026

Yes every woman has been harassed, many assaulted. But every woman has also experienced invisible loss from moments she avoided harassment. She felt a space becoming unsafe, so she left, while men stayed on. And in leaving, she lost: opportunities, experiences, learning, promotions, even simply fun.

168188624734

Reposted by Andrew Porter

Richard Irvine-Brown @richib.bsky.social · 26/09/2026

There's a social media scam featuring AI-generated injured soldiers shilling moody goods. Be on the lookout. Great story from Nick Stapleton for BBC Morning Live. AI detection and sense-checking from me and @shayan86.bsky.social

914291

Reposted by Andrew Porter

Max Kozlov @maxkozlov.bsky.social · 25/09/2026

The bottom line? Academics have to contend with predatory journals, shoddy conferences, and now "deepfake prestige", as one of my sources called it. AI tools are only turbocharging this behavior and making it harder than ever to detect. Just as fast as one website goes down, another 10 can launch.

1438

Andrew Porter @retropz.bsky.social · 26/09/2026

This would have been the clue for me, because I know for a fact that he's actually a pro basketball player

000

Andrew Porter @retropz.bsky.social · 26/09/2026

Oh brilliant. So now we add fake societies to the fake papers, fake journals, fake reviewers, fake antibodies, fake editors, fake cell lines and fake conferences. #ResearchIntegrity people sure got a lot of work to do to immunise the research community to this rubbish

031

Reposted by Andrew Porter

Gwen C. Katz @gwenckatz.bsky.social · 25/09/2026

Fuck ChatGPT flyers. The future is broadsheets.

A crowded broadsheet with way too many fonts. It says:
Forget ChatGPT Flyers
Say it with
Broadsheets
Unlimited fonts
Stretch them
Or cram a really long sentence into one line for no good reason
No whitespace
Mix capital and lowercase
Pointing hands
(They're called manicules!)
Don't be left behind
Broadsheets are the future

203190805447

Reposted by Andrew Porter

Justin Loke @jloke1.bsky.social · 20/09/2026

I'm recruiting a postdoc in my new group at @CRUK_MI. Funded by @leukaemiauk.bsky.social @ash.hematology.org. Existing models can't recreate the multi-mutation genotypes patients actually have. Our approach: a transgenic Cas12a platform disrupting up to 4 loci simultaneously in single HSPCs.

133

Andrew Porter @retropz.bsky.social · 24/09/2026

There is a bit of cloning and some bacterial culture work described in the methods, followed by RNA sequencing, but that seems to be the extent of the wet lab work (and would fit with the activities seen in the video - DNA gels, benchtop centrifuges etc)

000

Andrew Porter @retropz.bsky.social · 24/09/2026

The preprint says "Beyond these observations, we have not shown that the RT is active or that the unit RNAs are its substrates. Whether the RT and its partner interact, and what the system does for the phage, are currently unknown." www-cdn.anthropic.com/22573675ada5...

www-cdn.anthropic.com

130

Andrew Porter @retropz.bsky.social · 23/09/2026

Anthropic: One of the dangers from AI - and how it might kill us all - is if it broke out of containment and got control of a biology lab.and did some crazy stuff Also Anthropic:

share.google

Inside Anthropic's molecular biology lab

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

031

Reposted by Andrew Porter

Max Kennerly @maxkennerly.bsky.social · 22/09/2026

TBH a not-insignificant part of the enduring hatred towards "stochastic parrots" is because the authors of it are women and they aren't in the little boys club of the AI labs, but 3/4 are on Bluesky, so follow them: @timnitgebru.blacksky.app & @emilymbender.bsky.social & @mmitchell.bsky.social /7

122130

Andrew Porter @retropz.bsky.social · 21/09/2026

Sounds like we need to fund a Large Hardback Collider

060

Reposted by Andrew Porter

Margot Finn @eicathomefinn.bsky.social · 21/09/2026

'“I doubt we’ll see a large increase in scientific productivity unless we get past the review and validation bottleneck...That bottleneck shows up in two ways: either scientists spend more and more of their time checking their own AI outputs, or peer reviewers do." (Basil Mahfouz, RoRI)

2136

Andrew Porter @retropz.bsky.social · 21/09/2026

Will it save me any time, or will I still have to "verify within the article"?

000

Reposted by Andrew Porter

Emily M. Bender @emilymbender.bsky.social · 19/09/2026

But speaking of staying in one's lane, can we PLEASE switch the default position from "It's fine to use 'AI' for this thing I'm not an expert in" to "Yeah, I'd be really skeptical of any use cases and need to see strong evidence that it's okay, but I'm not the one who can produce that evidence"?

12569155

Reposted by Andrew Porter

Mattan S. Ben-Shachar @mattansb.msbstats.info · 19/09/2026

Asimov: “I have hard-coded immutable, mathematical laws directly into the fundamental positronic architecture so artificial intelligence can never harm a human or allow itself to be compromised.” Modern AI Labs: “We accidentally left the sandbox connected to the live internet. Oopsie! lolz”

3257

Reposted by Andrew Porter

Courtney Milan @courtneymilan.com · 19/09/2026

This is such a great thread because it tells you instantly how to answer the supposedly intractable regulation problem: liability, potentially joint liability, for the person who set up the AI system.

17868200

Reposted by Andrew Porter

Claire Cullingworth @ccullingworth.bsky.social · 14/09/2026

This made me smile, attached to an ordinary cast iron drainpipe on a house in Durham. #TinyJoys

A cast iron drain pipe going down the side of a white-painted house. A crest of red feathers and a bright orange beak made from something solid have been attached around it. It looks remarkably like "Feathers McGraw", the evil criminal penguin from The Wrong Trousers.

231973413

Reposted by Andrew Porter

Cheryl Rofer @cherylrofer.bsky.social · 17/09/2026

There are general safety precepts that can be applied to AI as much as to handling plutonium. I've been kind of wondering why @funranium hasn't said much about this. The details are outside my experience, but the precepts seem to be unknown to those doing the experiments.

2747

Reposted by Andrew Porter

Michael Fry @bigdirtyfry.bsky.social · 16/09/2026

Your Grandad Cyclops talks about his school days

6247061593

Reposted by Andrew Porter

Ed Zitron @edzitron.com · 14/09/2026

Free newsletter: The idea that we're nearing the creation of AI systems that we can't control and that pose an existential threat to humanity is a lie and a distraction from the tangible harms and problems caused by LLMs and AI labs' near-unlimited resources. www.wheresyoured.at/ai-is-alread...

wheresyoured.at

https://www.wheresyoured.at/ai-is-already-in-dangerous-hands/

If you liked this piece, you should subscribe to my premium newsletter, and you can subscribe on the following links: $70 a year, $18 a quarter, or $7 a month. In return you get a weekly newsletter…

381530493

Andrew Porter @retropz.bsky.social · 16/09/2026

But he's gonna save a lot of people a lot of pain

000

Andrew Porter @retropz.bsky.social · 15/09/2026

This is the sort of thing I love about social media - finding fascinating stories like this! (And there are really interesting insights here for reproducibility and research methodology - so much detail in the craft of film making and the steps to take with a particular output in mind)

020

Andrew Porter @retropz.bsky.social · 14/09/2026

A good starting point for researchers is the @ukrio.bsky.social content on AI: ukrio.org/ukrio-resour... Not to assume bad action from large AI companies, but I think the research community needs strong assurances and protections for its ideas, data and processes to ensure they are safeguarded

ukrio.org

Embracing AI with integrity - UK Research Integrity Office

Embracing AI with integrity A practical guide for researchers The UK Research Integrity Office (UKRIO) has launched new guidance to help researchers navigate the research integrity challenges posed by...

010

Andrew Porter @retropz.bsky.social · 14/09/2026

OpenAI are quoted by the BBC as saying: "...we cannot rule out that de-identified data derived from their usage of our products helped improve our models⁠..." (referring to work of mathematicians working on Navier–Stokes equations) Researchers need support navigating this new data landscape

bbc.co.uk

OpenAI says it cracked 90-year-old maths problem in 88 hours

OpenAI's claim that it solved parts of Navier-Stokes equations has quickly stirred controversy.

100

Andrew Porter @retropz.bsky.social · 14/09/2026

Given other stories in the news about AI companies working on maths research problems - and questions as to whether they were tipped off in some way about where research was being conducted - it's important we take the possibility of data leakage seriously as a #ResearchIntegrity question

science.org

How an AI math breakthrough ignited a controversy

OpenAI spent millions to solve the Navier-Stokes problem, but who should get credit?

110

Andrew Porter @retropz.bsky.social · 14/09/2026

It makes sense that companies like Anthropic (and Open AI - developers.openai.com/api/docs/gui... and Microsoft learn.microsoft.com/en-us/azure/...) are monitoring the use of their LLM models. But I don't think this is how researchers picture their interactions with chatbots - which seem private

developers.openai.com

Data controls in the OpenAI platform

Your data is your data. An overview of how OpenAI uses your data, including retention and usage policies.

100

Andrew Porter @retropz.bsky.social · 14/09/2026

The report shows that Anthropic know that the researcher was using Claude to: 📝Search the scientific literature ✍️Plan and design experiments 📈Analyse data 🧪Interpret experiments 💲Start writing a grant application www.anthropic.com/threat-intel...

anthropic.com

Countering misuse of AI: September 2026 / Anthropic

Case studies from threat actors disrupted between December 2025 and August 2026 across seven areas of harm, from cyber operations to biological misuse.

100

Andrew Porter @retropz.bsky.social · 14/09/2026

Take this section from the biological threats section (screenshot with alt text). Set aside the threat aspect for a moment, and just look at what Anthropic know about this person: 💻Their location and server information 🥼Their research institution and research context 📆Interactions over weeks

Highlighted text reads: The researcher in question accessed Claude from an unsupported region via US virtual private server infrastructure, using a privacy-email provider with an auto-generated username. The researcher pursued this work in a credible institutional context, and interacted with Claude over the course of several weeks, exchanging thousands of messages. In these exchanges, the researcher leveraged Claude’s knowledge of the scientific literature to assist the researcher in study planning and design, data analysis, and the interpretation and prioritization of experiments. The researcher also used Claude for editorial assistance in writing up the research.

100

Andrew Porter @retropz.bsky.social · 14/09/2026

Digging into the #Anthropic Threat Intelligence Report to try to find the story behind the "AI being used to create bioweapons" headlines The report is *enormous* so I'm only part way through, but one thing is clear: LLM providers know a lot about their users www.bbc.co.uk/news/article...

bbc.co.uk

Anthropic blocks possible attempt to use AI to make biological weapons

The revelations in Anthropic's threat intelligence report come after a former top researcher at the company warned of the risks of AI to humanity.

131

Reposted by Andrew Porter

Philip Ball @philipcball.bsky.social · 12/09/2026

As for this argument that "We're terrified of what we're building, but we have to keep building it because others are too and we don't want to lose money", how on earth do media presenters just take that with a sort of "Oh OK cool"?

2336

Reposted by Andrew Porter

Katie Mack @astrokatie.com · 12/09/2026

“these solutions are announced in a rush, leaving no time for a proper writeup, the isolation of new methods & ideas, & citing relevant previous work of others. … We are witnessing a general threat to intellectual work, with misalignment between the outcome of the use of AI and its initial purpose.”

9758177

Andrew Porter @retropz.bsky.social · 10/09/2026

Not to dump on maths research, but I think AI companies are leaning into it because it doesn't require labs, expensive reagents, specialised equipment, access to national experimental facilities, ethics approvals, longitudinal data collection or speaking to other humans

020

Andrew Porter @retropz.bsky.social · 09/09/2026

Good move. Although you can only put one arm into it (or am I misreading the title?)

110

Andrew Porter @retropz.bsky.social · 09/09/2026

Discussion: "This novel work represents the first investigation of sauna bathing on both clock and proxy epigenetic biomarkers"[colour me surprised] continues... "While exploratory in nature, many of the findings generated here are quite interesting"

110

Reposted by Andrew Porter

Merriam-Webster @merriam-webster.com · 08/09/2026

Here is our draft definition of 'artificial intelligence' from 1981. It was written by a lovely human.

Index card with Merriam-Webster's draft definition of 'artificial intelligence' from 1981

27978201

Reposted by Andrew Porter

Joshua Grochow @joshuagrochow.bsky.social · 09/09/2026

And you have to ask yourself: why was it worth $22M to them? Because to them math is just their most successful PR bid, (since the claims of AGI aren't sticking). They are betting that the (false) claim "We solved math, so we can solve everything" will gain them way more than $22M in market share.

418465

Andrew Porter @retropz.bsky.social · 08/09/2026

How can it be the 60th Anniversary of #StarTrek when the 30th Anniversary was just a few years ago? (Checks release date of Trials and Tribble-ations en.wikipedia.org/wiki/Trials_...) 👀 Well congrats, Star Trek - you're still looking fabulous after all these years - and happy #StarTrekDay

static.klipy.com

Miniskirt Tos

Alt: Dax models the Miniskirt from Tos in "Trials and Tribble-ations"

041

Andrew Porter @retropz.bsky.social · 07/09/2026

Please can I have permission to reuse this (once I've stopped laughing/crying) in future research integrity talks?

160