How artificial intelligence will actually work...

One good reason not to use facial recognition: #privacy breaches of data that cannot be modified (faces, unlike passwords, cannot be changed). Other good reasons include preserving freedom and democracy.
“As always, #surveillance isn’t safety.”

#AIEthics
wired.com/story/outabox-…

We are hiring two PhD researchers for my project PLATFORM WARS UAntwerpen: (1) On military tech and changing socio-technical practices of warfighting (IR, STS, Anthropology); (2) On the regulatory role of hi-tech firms (IR and legal studies). ⏰Deadline for applying: 10 July.

Scarlett Johansson is right: AI companies must be more transparent on.ft.com/3WSRtD7 | opinion

Not really the most important thing today but I chatted with Dan Milmo of The Guardian about how the UK as a whole seems to be paying too little attention to the Chinese cyber threat to civilian infrastructure and too much to standard gov’t espionage 1/2

theguardian.com/uk-news/articl…

NEW: In an internal Meta threat assessment I obtained, WhatsApp engineers warned users are vulnerable to government spying that unmasks who's talking to who. Employees later speculated Israel may be exploiting this to target and kill across the Gaza Strip theintercept.com/2024/05/22/wha…

Last week, I went to an “AI Expo” that was put on by Eric Schmidt’s think tank and funded by Palantir. It was incredibly bleak and surreal. For The Guardian, I wrote about my experience, and the people I met:
theguardian.com/technology/art…

#Canada joins international security partners in release of advisory, guidance on growing #CyberSecurity threat to civil society.

Last week at our annual academic conference, leading researchers from diverse fields gathered to tackle the complexities of AI alignment and to better understand the social impacts of data-driven technologies. Read a summary of the conference on our blog: uoft.me/au-

“The more Russia sees itself as falling behind the West in high-tech development, the more vulnerable it may perceive itself to be, which may lead Moscow to accept greater risk in the way that it deploys #AI to keep apace.”

cnas.org/publications/r…

Today is pub day for a big report I wrote for ICRC on AI in conflict.

Humans who use AI in war should always be able to make an appropriate decision, regardless of whether the computer is right or wrong. The report shows why that’s easier said than done.
shop.icrc.org/decisions-deci…

If you are 🇨🇦, do vote for the NHL Willie O’Ree Community Hero Award. Allen Hierlihy is so deserving of this award! nhl.com/community/will…

Here are some thoughts from me about the rush to deploy AI and increasingly autonomous systems by militaries. One point that I added in the longer conversation is that in densely populated urban areas these risks are acute. cbc.ca/news/politics/…

'Chatbots can accurately infer an alarming amount of personal information about users—including their race, location, occupation, and more—from conversations that appear innocuous.'
#privacy #LLMs #AIEthics

wired.com/story/ai-chatb…

Microsoft created an air-gapped ChatGPT for spies to use. Most generative AIs use the cloud, but US intel community wanted ability to run queries on classified info and not have it leak to the cloud or get hacked. 'it’s on a special network that’s only accessible by the US

Call for papers for a workshop on #EmergingTech in international security, specifically for early career researchers - submit your abstracts and come visit us CWS - War Studies The AutoNorms Project

It was wonderful to discuss the ethics of purposely harmful AI systems, both within and outside the context of warfare, with Branka Marijan and Leah West. Huge shoutout to Jamie Duncan for the fantastic moderation and to the audience for the weighty questions!

💡 We're kicking off Absolutely Interdisciplinary 2024 with a fantastic series of talks at our graduate workshop, 'Interdisciplinary Dialogues on AI.'

Learn more and register for the afternoon sessions: uoft.me/sri-graduate-w…