Tim de Sousa's Avatar

Tim de Sousa

@timdesousa.bsky.social

Privacy, Information Governance and Tech Ethics wonk, based in Sydney Australia. Mediocre ukulele player. Slightly snobbish about coffee.

204 Followers  |  416 Following  |  432 Posts  |  Joined: 19.11.2024  |  2.2863

Latest posts by timdesousa.bsky.social on Bluesky

"are you enjoying duo mobile" does a hamburger enjoy being made of quarks. does a fish enjoy linear time. does the mountain enjoy the first taste of a cup of hot chocolate when you get back to the ski lodge. your question means nothing to me. i couldn't enjoy duo mobile even if i tried

03.12.2025 19:36 β€” πŸ‘ 1456    πŸ” 314    πŸ’¬ 20    πŸ“Œ 14

This Pluribus ad on a fridge caused a schizophrenic woman named Carol to be hospitalized.

04.12.2025 03:08 β€” πŸ‘ 6587    πŸ” 2856    πŸ’¬ 148    πŸ“Œ 299

I deeply hope that this article is inaccurate.

04.12.2025 03:57 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Beyond that - the data used to create this system is not likely to be high quality. There have been numerous complaints about the administration and inaccuracy of the NDIS process. To automate based on that would be deeply irresponsible.

04.12.2025 03:57 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Preview
A Trust Framework for Government Use of Artificial Intelligence and Automated Decision Making This paper identifies the current challenges of the mechanisation, digitisation and automation of public sector systems and processes, and proposes a modern and practical framework to ensure and assur...

This fails 3 out of 4 of these requirements! It doesn't have to be this way. This is an avoidable tragedy. arxiv.org/abs/2208.10087

04.12.2025 03:56 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Beyond that, nothing described here suggests that this systems will achieve the elements of a trustworthy automated decision-making systems.
1. TRANSPARENT rules
2. TRACEABLE application of rules to evidence
3. ACCOUNTABILTY - the decision of the machine is a decision of the agency
4. APPEALABLE

04.12.2025 03:55 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

For one, you need to provide reasons for your decision, and you need to be able to appeal!

04.12.2025 03:53 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

This seems unlikely to meet basic requirements for government administrative decisions or best practices for automated decisions.

04.12.2025 03:51 β€” πŸ‘ 2    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

No to my knowledge.

04.12.2025 03:35 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

This is a extremely poor idea: "According to revelations in Guardian Australia this week, there will be no safety net – no scope for either the NDIA delegate or the administrative appeals tribunal to amend the budget. In other words, the computer will have the final say."

04.12.2025 02:17 β€” πŸ‘ 10    πŸ” 10    πŸ’¬ 2    πŸ“Œ 0
Preview
Government to target 'dark patterns' behind hidden fees and subscription traps

πŸ“’ The government wants to outlaw drip pricing and subscription traps, targeting manipulative online tactics that cost Australians millions.

30.11.2025 23:04 β€” πŸ‘ 7    πŸ” 8    πŸ’¬ 0    πŸ“Œ 0

The time is upon us once more, godspeed all.

30.11.2025 22:55 β€” πŸ‘ 4    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
Use of AI β€˜tech stacks’ driving demand for lawyers The more artificial intelligence is used within a law firm, the more professionals are needed to vet the technology’s outputs.

Loving that AI productivity boost!

"the more artificial intelligence is used within a law firm, the more lawyers are needed to vet the technology’s outputs."
www.afr.com/companies/pr...

30.11.2025 22:50 β€” πŸ‘ 280    πŸ” 73    πŸ’¬ 21    πŸ“Œ 23

Do this even if you are old.

28.11.2025 04:50 β€” πŸ‘ 4    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

I'm old.

If you're not, listen to this.

Get a guitar or some drums, or a trumpet. Pick up a brush, a pencil, some clay or a welder.

Just make shit. It doesn't have to be good (eventually it will be) just make music, art or write or whatever.

It's literally never been more important.

27.11.2025 04:21 β€” πŸ‘ 8525    πŸ” 3130    πŸ’¬ 146    πŸ“Œ 156

Mood

28.11.2025 02:00 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Shocked Stone headgehog

Shocked Stone headgehog

Part 218 of 200 in historically interesting things to inspire your ttrpg dnd

Ceramic hedgehog dated to the Neolithic period found at the Hamin archaeological site near Shebotu Town in Inner Mongolia, China

22.11.2025 03:14 β€” πŸ‘ 802    πŸ” 226    πŸ’¬ 17    πŸ“Œ 82
Video thumbnail

Good morning to Brazilian reporter Manuela Borges, who’s been waiting eleven years for this petty moment. ❀️ πŸ‡§πŸ‡·

26.11.2025 13:04 β€” πŸ‘ 56264    πŸ” 14981    πŸ’¬ 832    πŸ“Œ 1976

This is extremely flattering, Mike! As I am a rank amateur, I don't have a website or an online shop or anything. But DM me your email address and I will send you a high res scan.

25.11.2025 22:11 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Been a bit time poor lately so had to do this in pieces over a few days. I think they’re more of a classic 90s boy band rather than a goth band.

25.11.2025 09:16 β€” πŸ‘ 63    πŸ” 10    πŸ’¬ 0    πŸ“Œ 1
Preview
Australia to establish AI safety institute The Albanese government will establish an AI Safety Institute to assess the risks from emerging artificial intelligence systems and help safeguard Australians from potential harms.Β Industry minister T...

Looks like we're getting an AI safety institute. Sensible.

www.innovationaus.com/australia-to...

24.11.2025 22:37 β€” πŸ‘ 11    πŸ” 5    πŸ’¬ 3    πŸ“Œ 2
Post image

Privacy and AI Deregulation Will Be Bad for Innovation - my latest Substack
open.substack.com/pub/danielso...

23.11.2025 19:42 β€” πŸ‘ 10    πŸ” 5    πŸ’¬ 1    πŸ“Œ 0

Wow, they’re gorgeous! And what a perfect yin yang!

22.11.2025 12:55 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Was tickled by these two

22.11.2025 06:35 β€” πŸ‘ 5    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

If you need me, I’ll just be crumbling into dust like that guy at the end of Indiana Jones and the Last Crusade.

21.11.2025 22:28 β€” πŸ‘ 9    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0

Relying on ChatGPT to teach you about a topic leaves you with shallower knowledge than Googling and reading about it, according to new research that compared what more than 10,000 people knew after using one method or the other.

Shared by @gizmodo.com: buff.ly/yAAHtHq

21.11.2025 11:48 β€” πŸ‘ 3283    πŸ” 1256    πŸ’¬ 79    πŸ“Œ 388

Time to move past the neverending trial stage. Yes, basic income reduces homelessness.

21.11.2025 04:30 β€” πŸ‘ 556    πŸ” 193    πŸ’¬ 1    πŸ“Œ 4

This actually brightened my day. You can defeat guardrails on LLM models by making your request in poetry.

*cracks knuckles, climbs up library ladder, starts browsing*

oh it's on now

21.11.2025 04:06 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Paying a ransomware demand makes businesses more likely to be targeted again despite the criminals’ assurances to the contrary, according to new Australian Institute of Criminology (AIC) research that found Australian SMEs receiving multiple ransomware demands www.aic.gov.au/sites/defaul...

21.11.2025 00:15 β€” πŸ‘ 8    πŸ” 8    πŸ’¬ 0    πŸ“Œ 0
Preview
New laws could stop bosses using AI to set rosters, techies warn Technology developers and online platforms are alarmed over proposed NSW laws regulating artificial intelligence, warning they risk the uptake of new technologies.

New laws could stop bosses using AI to set rosters, says Tech Council of Australia www.afr.com/work-and-car...

21.11.2025 00:17 β€” πŸ‘ 29    πŸ” 7    πŸ’¬ 2    πŸ“Œ 1

@timdesousa is following 20 prominent accounts