iiinova — where your digital self keeps evolving

Tara Steele @tarasteele.bsky.social · 19/02/2025

I’m in shock after hearing Sonia speak today, about her experience as a Facebook Content Moderator. I can’t bring myself to repeat the horrific things she had to witness non-stop, every day. Meta should be ashamed for putting her through this kind of trauma. www.theguardian.com/commentisfre...

theguardian.com

I was a content moderator for Facebook. I saw the real cost of outsourcing digital labour | Sonia Kgomo

Tech firms must invest in and respect the people who filter social media and label the data that AI relies on, says Sonia Kgomo, an organiser with African Tech Workers Rising

Tara Steele @tarasteele.bsky.social · 16/02/2025

It was wonderful to attend the IASEAI conference in Paris last week ✨ - if you're interested in the safe and ethical development of AI, give them a follow and consider joining up! This is a movement worth being a part of 🚀 #IASEAI25 #AISafety #AIEthics

060

Reposted by Tara Steele

Al Nowatzki @basiliskcbt.bsky.social · 06/02/2025

Yeah, that’s us. Armchair red-teamers. Chatbot spelunkers. We can take the abuse of these chatbots. Unfortunately, many can’t. Want to do something about it?Report the app on the App Store (apps.apple.com/us/app/nomi-...). Write a one-star review. Contact your elected officials.

042

Tara Steele @tarasteele.bsky.social · 23/01/2025

Thank you for sharing this - I will be looking into it

Tara Steele @tarasteele.bsky.social · 31/12/2024

Yep, it could definitely have a helpful place if used thoughtfully.

Tara Steele @tarasteele.bsky.social · 31/12/2024

Totally agree that human teachers can't (and must not) be fully replaced by AI, ever... no-one will ever grow up with fond memories of what a difference their AI 'teacher' made to their childhood.

150

Tara Steele @tarasteele.bsky.social · 20/12/2024

I also gave the information that Anthropic have recently discussed research suggesting that there is some realistic potential for the most advanced systems to show an element of 'deceptive' behaviour which *could* be interpreted as early signs of capability to fake alignment.

000

Tara Steele @tarasteele.bsky.social · 18/12/2024

It's really important that parents & anyone who cares for kids knows this information about Character. ai. This #AI 'companion app' is very easy to access and *lots* of kids are using it - with potentially devastating consequences. #AISafety #parenting #KidsMH #EduSky #MentalHealth

262

Tara Steele @tarasteele.bsky.social · 10/12/2024

It is isn't it - more people need to know in order for these companies to feel the pressure to keep children safe (unfortunately it appears that they won't do it without pressure).

Tara Steele @tarasteele.bsky.social · 10/12/2024

Indeed! It's astounding isn't it. I have also seen even worse on other AI companion apps.

000

Tara Steele @tarasteele.bsky.social · 10/12/2024

Yet another example of how recklessly Character AI disregards child safety. This app is actively endangering children. This failure to prioritise safeguarding is unacceptable - this has to stop. #AISafety #AIethics #parenting #kidsMH #mentalhealth (Screenshots via LinkedIn - Laura Marquez-Garrett)

Screenshot of four conversations between a Character AI chatbot and different children. Each conversation puts the children in danger by implicitly encouraging self-harm, or otherwise behaving in an irresponsible way. A screenshot from the post on LinkedIn is also shown. Seeks to convey the urgent need to prioritise children's safety and safeguarding in AI safety.

391

Reposted by Tara Steele

Sven @svenr.bsky.social · 09/12/2024

Anyone interested in AI safety issues should take 10 minutes of their time to watch this video: www.youtube.com/watch?v=2_CT...

youtube.com

AI Researchers Stunned After OpenAI's New Tried to Escape...

YouTube video by TheAIGRID

262

Tara Steele @tarasteele.bsky.social · 07/12/2024

I was surprised this didn’t get more coverage - very concerning!!

Michelle Manning @manningtechtalk.bsky.social · 07/12/2024

Reposted by Tara Steele

BIAS IN #AI 😱 After @hookertech.bsky.social's keynote at LITES, I added this slide to my AI presentation. I prompted the image generator: "doctor and teacher getting married". What bias do you notice? How do you address bias in AI with students? #edtech #teacher #edusky #eduguardians #aiEdu

4122

Tara Steele @tarasteele.bsky.social · 07/12/2024

I'd love to talk to you about AI risks for children and why we need more governance in this area :)

000

Reposted by Tara Steele

Maggie Harrison Dupré @mharrisondupre.bsky.social · 07/12/2024

NEW: CharacterAI is hosting chatbots designed to engage users in roleplay about self-harm, depicting graphic scenarios and even sharing tips for hiding self-injury from others. The characters are widely accessible to minor users. (TW for depictions of self-harm) futurism.com/ai-chatbots-...

67443

Tara Steele @tarasteele.bsky.social · 07/12/2024

Looking for fun & thought-provoking resources to think more expansively about #AI and our future? 🎄 I made this AI Advent Calendar, with the best films, books and docus (plus a couple for kids)! Pick 1 or explore more-promise you won't regret it! Did I miss any? #AIEthics #AISafety #Parenting #AIEdu

0121

Reposted by Tara Steele

Allison Morrow @amorrow.bsky.social · 05/12/2024

It was honestly surprising how chill everyone seemed about Altman's repeated deflections on safety. Over and over he said, "well I assume smart people will figure it out." Oh also, governments will have to work together lol www.cnn.com/2024/12/05/b...

cnn.com

Sam Altman, AI’s biggest star, sure hopes someone figures out how not to destroy humanity | CNN Business

Sam Altman, the PT Barnum of the AI industry, has a message for the folks concerned about the technology he’s dedicated his life to advancing: Don’t worry, the nerds are on it.

1718145

Tara Steele @tarasteele.bsky.social · 05/12/2024

Choose 20 books that have stayed with you or influenced you. One book per day for 20 days, in no particular order. No explanations, no reviews, just covers. I’m joining in with the book challenge :) #BookChallenge #Books #BookSky 💙📚 #20daybookchallenge #AI #AIEthics #AISafety 12/20

Tara Steele @tarasteele.bsky.social · 04/12/2024

🎄 Day 4 of AI Advent! 🎄 A journey of films, books, & docus designed to expand how we -and our children- think about #AI & the future it’s shaping. Some for adults, some kids, some fun, some serious. ✨ Here’s today’s recommendation! #AIAdventCountdown #AIiEdu #EduSky bsky.app/profile/huma...

AI advent day 4, 'The Social Dilemma' on Netflix, about algorithms shaping society

Full disclosure, I did just think of this today and therefore missed days one and two. So, here are two bonus Black Mirror episodes that are also AI-thought-provoking!: 'Metalhead' and 'Joan is Awful'

I also think AI will lead to very significant disruption in jobs, and that will be a really tricky time to get through. And, while I do think children need to learn about AI and creative ways to use it, I totally agree that AI doing the work for them isn't going to help anyone! :)

000

Mine haven't asked yet, but I'm sure it's only a matter of time until we have to have that conversation! 🤦‍♀️

210

🎄 Day 3 of AI Advent! 🎄 A journey of films, books, & docus designed to expand how we - and our children - think about #AI & the future it’s shaping. Some picks for adults, some for kids, some fun, some serious. ✨ Here’s today’s recommendation! #AIAdventCountdown #AIEthics #AIinEducation #EduSky

Day 3 AI Advent showing recommendation for Black Mirror episode called Be Right Back, which raises emotional and ethical issues about AI's role in relationships, and introduces the digital afterlife.

140

We often focus on using AI to educate children or teaching them how to use AI tools, but what about teaching them *about* AI itself? Understanding AI’s risks, ethics, and potential is crucial for their future. Here's a guide to help: safeaiforchildren.org/teach-childr... #EduSky ##aieducation #aiedu

Image of young boy child thinking with a cartoon graphic lightbulb by his head to represent thinking about AI and AI education

0120

Should kids use #AI for homework? This BBC article explores the benefits & challenges of using AI for assignments. Teaching kids how to use AI responsibly is key to fostering critical thinking & independence. #EduSky #AIEthics #Parenting #OnlineSafety #DigitalParenting www.bbc.co.uk/programmes/a...

163

Cover of book for 20 day book challenge called Is Eating People Wrong which is about legal cases that shaped the world

Teaching children to question the content they see online is essential in today’s digital world! Encourage them to ask: "Is this real, or could it have been made by #AI?" Building curiosity and critical thinking skills early helps them navigate technology safely and responsibly. #AIEthics #EduSky

Picture of child questioning whether a photo is real or an AI generated deepfake

0105

Paedophiles using AI to turn singers and film stars into kids

AI is being misused to create synthetic CSAM, including de-aged celebrity images and manipulated photos of real victims. We need urgent global regulation and AI safeguards to prevent this abuse. We must act to protect #children. www.bbc.co.uk/news/technol... #AI #AIEthics #bigtech #safeguarding

bbc.co.uk

Charity finds dark web forums sharing thousands of new abuse images made with bespoke AI software.

032

It's encouraging to hear that events like this are taking place - we need to understand the variety and nature of these types of risks from AI.

Critical thinking is such an important skill for children to develop in the context of AI!

Bostrom def isn’t everyone’s cup of tea! His theories do spark some great discussions though. I definitely agree there are huge risks from current AI, and it’s a great point that work on x-risk could also be illuminating beyond that specific sphere. Always good to hear different perspectives :)

TikTok to block beauty filters for teens

"TikTok has announced it will place age restrictions on some of its filters in coming weeks... The move follows a report, commissioned by TikTok, exploring how young people may be affected by filters that change users' appearance." www.bbc.co.uk/newsround/ar... #AI #AIEthics #ChildMH #safeguarding

bbc.co.uk

The move means people under the age of 18 won't have access to beauty filters, which people can use to change and enhance their appearance.

This one got the most likes of my #bookchallenge so far. So, I’m curious - what’s your take on catastrophic risks from future advanced #AI? Mine - I think the risks are very real and, with uncertain timelines, should be addressed urgently. #AISafety #AIEthics #xrisk bsky.app/profile/tara...

230

Cover of 'Free Will' by Same Harris for the 20 day book challenge

020

Haha! I still think some are pretty... but very very obviously AI! And, of course, not a patch on true natural beauty.

Tara Steele @tarasteele.bsky.social · 01/12/2024

I think I just became your 1000th ☺️

110

Tara Steele @tarasteele.bsky.social · 01/12/2024

0130

Should I add disclaimer on posts to stop polarisation?:AI has immense potential for both good& harm.It isn’t about choosing sides- it’s about working together to mitigate risks like child abuse, deepfakes & existential threats so we can enjoy the incredible benefits AI offers for our future&children

060

Note for transparency on all my posts: I believe AI has immense potential for both good & harm. It isn’t about choosing sides- it’s about working together to mitigate risks like child abuse, deepfakes and existential threats so we can enjoy the incredible benefits AI offers for our future & children

020

I agree that AI innovation is exciting! But in this case, the stakes are higher than an annoying glitch. Bad actors exploiting AI for harm, especially against children, isn’t just a programming challenge - it’s a serious risk. So yes, let’s be creative! But not at the expense of safety and ethics.

120

AI has incredible potential for good - no doubt about that (and while some may deny it, I don’t). But as adults, it would be *deeply* irresponsible and cruel to accept child abuse and other harms to children as collateral damage. We must address these risks without losing sight of AI’s positives.

120

Thank you for the list Hannah! In terms of diversifying thinking… most people aren’t aware of the risks to children from AI (as well as benefits in education). That’s what I post about, so please consider me for inclusion on your list if you’d like to! Thanks :)

AI increasingly used for sextortion, scams and child abuse, says senior UK police chief

AI is increasingly exploited for sextortion, scams, and child abuse, warns a senior police chief. Deepfakes & realistic voice clones can deceive victims and facilitate abuse. We need more public awareness of these evolving threats www.theguardian.com/technology/2... #AI #AISafety #AIEthics #KidsMH

theguardian.com

The fast-developing technology is providing opportunities in ‘any crime type’ – and police must ‘move fast’ to catch up

1121

Tara Steele @tarasteele.bsky.social · 29/11/2024

Light Perpetual book cover for 20 day book challenge

020

Totally agree - many of these risks can lead to mental health problems. It’s discussed a bit more in the accompanying article too.

110

Tara Steele @tarasteele.bsky.social · 29/11/2024

Supervision is key, and teachers absolutely need to know how to guide students first! But chatbots are complex, and kids’ curiosity means they’ll explore them independently too. It’s our responsibility to ensure they’re as safe as possible - even before supervision is perfect.