Supporting teams at the Stanford Internet Observatory (@stanfordio), Election Integrity Partnership (@EI_Partnership) and https://t.co/gNIyykKdSu.
@alex@cybervillains.com
16 subscribers
Apr 20, 2023 • 4 tweets • 1 min read
In a world where the rapid advance of open-source generative AI is leading to a tidal wave of near-zero-cost BS flooding every text, image and video channel, eliminating account verification is a massive, historical mistake.
Will at least be fun to watch.
90% of mass media stories about AI safety are focused on a handful of big companies and the possibility of making LLMs say disagreeable things. There has been very little attention paid to the dark corners where groups are running models locally to cause actual harm.
Jan 25, 2023 • 4 tweets • 2 min read
Now that Twitter is censoring users on behalf of the Modi government in India, I'm reupping this October prediction that Musk's international entanglements would be the Achilles heel of Twitter.
Twitter should release all communications with governments regarding censorship.
Here is a conversation with @evelyndouek on what happened this week, and how this is a major departure from the unwritten policy of major US platforms to resist censorship requests that strike directly at core political freedoms.
Hopefully that button doesn’t actually work for them; Twitter has put a lot of work into internal access controls. If those two have been provisioned DM access, I expect that’s, at a minimum, an FTC reportable security incident.
Dec 3, 2022 • 18 tweets • 8 min read
Let's steelman @elonmusk's concern about political influence on the content decisions of major platforms. and come up with practical steps Musk could take to set a new standard:
1) Commit to releasing all communications by global political actors related to content moderation.
This would include all communications with government affairs teams, official requests from "internet referral units" such as those run by the EU and Israel, and out-of-band comms with executives such as Musk himself. You would want to include governments, parties and candidates.
Dec 3, 2022 • 8 tweets • 3 min read
This thread confirms what I knew! Excellent!
Twitter's reaction to 2016 included policies for both on-platform trolling and hack-and-leak. The latter lead to a big overreach that had nothing to do with the deep state or @EI_Partnership. A bunch of execs, of course, argued.
It's clearly not Twitter's responsibility to prevent the US media from publishing hacked, faked or stolen information. Had a great conversation on how platform's shouldn't overreach beyond their borders with the @TWiT team this week:
