Jase Gehring
@skyjase.bsky.social
scientist at UC Berkeley inventing advanced genomic technologies lover of molecules, user of computers https://scholar.google.com/citations?user=63ZRebIAAAAJ&hl=en
this has long bothered me about climate change discourse, and it's being repeated with AI. setting the bar at "human extinction" is insane. it's justifies all kinds of harmful choices.
Ketan Joshi@ketanjoshi.co · 8h agoAs I warned you: the superstar media darling AI quitter is explicitly agreeing with Trump and claiming climate threats are "overplayed". Effective altruism / rationalism is a cult of broken priorities that denies the threat of fossil fuels. This is why Anthropic loves burning gas for power.
wow this is a fantastic discovery. i thought natural DNA phosphorothioates were crazy, this is another level
Cell - a Cell Press journal@cp-cell.bsky.social · 3h agoNow online! Phosphate backbone epitranscriptomics: Discovery of natural RNA phosphorothioates and their writer machinery
yes it is very strange to see a technological revolution in progress, one driven by capitalist incentive that is expected to increase GDP, and say "nah, looks expensive"
Vladimir Salnikov@v4ldelund.bsky.social · 22h agoAfaik at this point EU had two solutions : 1. Treat this situation like WW2^covid, mobilize resources and try to catch up at least with China on AI & robotics 2. Do nothing And this is essentially the second option on that screenshot
This article is based on a single acquisition. It’s trend-casting
Nature Biotechnology@natbiotech.nature.com · 2d agoA new class of biotech buyer is emerging, as frontier AI and big tech companies acquire early-stage biology foundational platforms. What are the potential implications for founders, investors and the pharmaceutical industry? www.nature.com/articles/s41... rdcu.be/URMEtSZqA84a
even funnier, Douthat is probably the most anti-science and anti-evidence reporter in mainstream media. he thinks you can believe whatever you want as long as it's for religious (preferably catholic) reasons. perfect person to inform the american people on an active scientific revolution
60 Minutes@60minutes.bsky.social · 4d agoNew correspondent Ross Douthat says "no story looms larger than the latest developments in artificial intelligence." The technology and its impacts will be a major focus of 60 Minutes reporting.
open source is a response to centralization of power, not safety concerns. open weight models are far more dangerous at equivalent capability levels. as capabilities increase, the case for open weights get weaker. this is a problem. this sucks.
Literally just enforce existing law
Elise Thomas@elisethomas.bsky.social · 5d agoI think there's actually quite an easy and obvious way to solve the collective action problem of how to make the AI companies stop conducting risky research, and it's to make it very clear that they're legally liable for anything their models do. www.politico.com/news/2026/09...
This isn’t what they are saying or thinking. The industry is large enough and sufficiently capitalized to keep making progress. Yes it would be convenient if AI progress became incremental but with unlimited resources that is not likely
Andy Famiglietti@afamiglietti.org · 5d agoOk, I have been batting around the idea that they know their incremental improvements in the pipeline for the foreseeable future, while powerful upgrades to a powerful technology, are not the God in a bottle they promised investors and they want to hide behind safety as an out and maybe I am right?
Something I’m not seeing discussed is that OpenAI and Anthropic are not trying to build adversarial AI. It’s clear that anyone (most likely a state actor) who wanted to build a malicious AI system could make something much more dangerous
you will notice how Tao's openness toward AI use lends him a high degree of credibility even beyond his scientific expertise
Brandon Bishop@brandontbishop.bsky.social · 6d ago"the push by AI companies to solve mathematical problems as a benchmark is detrimental to the science of mathematics, and to the mathematical community. [...] We see these as part of broader alignment issues impacting other scientific and creative professions, as well as the whole of society."
anthropic's systems are being used for ethnically targeted attacks, mass surveillance, election interference, and autonomous weapons development. they admit to this in Russia, China, and elsewhere, and there is every reason to believe AI is used for the same activities inside the US government.
Anil Dash@anildash.com · 6d agoWe've had a million news cycles about the freakout from one guy who worked at Anthropic for 5 minutes saying we're all gonna die, but almost *nothing* about an actual multi-millionaire tech leader calling for an ethnic cleansing against the Roma people — an act which has happened in living memory!
another entry in "things that could be good if we has any confidence they were being done in good faith"
Colin Carlson@colincarlson.bsky.social · 7d agohaha what da fuck nsf-gov-resources.nsf.gov/files/Statem...
anthropic reports prompts for claude to write grant proposals related to virus research. part of the reason these are so lame is because claude isn't very helpful in a biological laboratory but check out how AI is supercharging weapons development and attacks on elections and disfavored groups
Jase Gehring@skyjase.bsky.social · 7d agohere is anthropic's report. the bio stuff is a yawn but holy shit the conventional weapons, cyberattacks, and social engineering case studies are insane. it's crazy the NYT article is about 'bioweapons' scarecrows when the report is packed with large-scale state-sponsored malicious use of AI
here is anthropic's report. the bio stuff is a yawn but holy shit the conventional weapons, cyberattacks, and social engineering case studies are insane. it's crazy the NYT article is about 'bioweapons' scarecrows when the report is packed with large-scale state-sponsored malicious use of AI
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.
anthropic.com
Jase Gehring@skyjase.bsky.social · 7d agoI’m not buying this without an independent report or analysis. I’ll update my thinking based on new evidence but “Anthropic said people used forbidden prompts“ has no informative value
I’m not buying this without an independent report or analysis. I’ll update my thinking based on new evidence but “Anthropic said people used forbidden prompts“ has no informative value
Richard Sever@richardsever.bsky.social · last wk."chilling examples of state-sponsored biological weapons developers tapping into the rapidly advancing capabilities of leading AI models" www.nytimes.com/2026/09/10/u...
I am again reminded of my experience from January of this year where Claude "invented" a new attention mechanism to incorporate mass and spatial relationships from mass spec data into deep models for de novo peptide sequencing. I was exploring new transformer architectures for this task, and
Paul Francois@pfrancois.bsky.social · last wk."The posted AI verification of Anthropic’s much-touted Jacobian conjecture (...) contains notational overlap with that work-in-progress, indicating that (...) the models can be so sensitive to previously-seen works-in-progress that even notation can leak through."
human health/longevity is used as one of the main reasons to pursue AI while human well-being/prosperity is assumed to be under threat. never count on rich people to bring about an equitable future
Latest #CDlab paper on BioRxiv: www.biorxiv.org/content/10.6... 1st author Eva Bertosin developed huge DNA origami nanopores that can dilate and constrict in size. With diameters up to 87 nm, they can insert into lipid membranes, showing selective transport when filled with nuclear pore proteins.
https://www.biorxiv.org/content/10.64898/2026.09.03.749129v1
biorxiv.org
Coders are saying LLMs are good enough for anything. They’re tackling historic math problems. In physical sciences LLMs haven’t changed research much at all
Been dreaming of giant sections like this for years. Honestly since high school. Whenever I look at a leaf this is what I want to see
Keri Martinowich@martinowk.bsky.social · last wk.Early data peek at Illumina's new StrataMap spatial transcriptomics cellular resolution platform. Huge capture area (7.5 cm2) - we profiled human 🧠 at level of thalamus - totally 🤩. Kudos to our teams for amazing work w/ nerve wracking sectioning & direct placement of this VERY large tissue section!
