Every safeguard DeepMind started with is gone:
2014: independent ethics board, reportedly a condition of the sale
2015: one informal meeting — then abandoned
2018: AI Principles exclude weapons & surveillance
2025: exclusions dropped
2026: Pentagon contract
Not one survived.
@Cyberleagle We really are moving towards a nanny state on this issue.
And of course, young people would never think of nor find ways to circumvent such ‘voluntary’ bans … & as parents, we felt a responsibility to limit access to a device & not allow it in our kids bedroom at night
In addition to the DPIA Microsoft 365 Copilot (see https://t.co/h0762Dj6r0, here's the Dutch DPA's Opinion pursuant to Article 36(2) of the GDPR regarding the proposed processing of Microsoft 365 Copilot by the Municipality of Haarlemmermeer https://t.co/x2CzeOXJQg
In addition to the DPIA Microsoft 365 Copilot (see https://t.co/h0762Dj6r0, here's the Dutch DPA's Opinion pursuant to Article 36(2) of the GDPR regarding the proposed processing of Microsoft 365 Copilot by the Municipality of Haarlemmermeer https://t.co/x2CzeOXJQg
The reported contract does not exclude mass surveillance, and it keeps paths open that could extend to autonomous policing.
These don't defend us against foreign adversaries. They shift power from citizens toward the state, in ways that are very hard to reverse.
I work at Google DeepMind. This won't make me popular. But it's all public reporting:
2014: DeepMind reportedly sold to Google on conditions: no military use, independent oversight
2026: a Pentagon contract for "any lawful government purpose"
Not one safeguard survived intact
This thread, absolutely.
Well-intended folk like @Livingstone_S & Baronesses Kidron & @FloellaBenjamin may demand in-app parental timelocks (etc) but there is no feasible way to provide them except "app-by-app" — which will be deemed "too complex/fiddly to use" by selfsame folk.
This thread, absolutely.
Well-intended folk like @Livingstone_S & Baronesses Kidron & @FloellaBenjamin may demand in-app parental timelocks (etc) but there is no feasible way to provide them except "app-by-app" — which will be deemed "too complex/fiddly to use" by selfsame folk.
Unused ventilators bought for £135m during Covid auctioned off for less than £1.5m
Millions of pounds of taxpayers’ money lost following government’s ‘ventilator challenge’, reveals FoI request
✍️ Nick Sommerlad
https://t.co/4RTfhZg4r0
“This was the plan all along” in Australia's Minimum Age #SocialMediaBan
https://t.co/SPJ5opvXNh
Expanding on Haidt's 11/07 tweet;
-making a market over 10 years;
-mandatory AV via govt-issued ID is undemocratic;
-starting softly was deliberate towards the final destination.
Education should not resemble a high-surveillance environment.
Privacy International is documenting the use of facial recognition in schools and universities. We need your input.
https://t.co/giyTkFusJV
A quick follow up on this audit.
cc @louismosley.
The Financial Times has found that the data behind NHS England's headline claims about Palantir's technology contains a "litany of errors". Four hospital trusts have confirmed the errors. The FT found trusts had reported discharge delays dropping from hundreds or thousands to zero overnight before spiking back up.
As the FT puts it: a scenario that would be "highly improbable". Charles Tallack, former head of operational research and evaluation at NHS England, said the evidence now looks "increasingly flimsy". The dataset, he said, "may be suitable for day-to-day management purposes, but not for evaluation". NHS England's own website admits there are only "minimal" quality checks on the data.
NHS England chief executive Sir Jim Mackey told Parliament's health and social care committee this week that he had personally challenged whether assessments of the platform's benefits "have been objective and can be fully stood up if challenged". He said an "objective" review would be "helpful and necessary".
The break clause decision is next spring. It's obvious how it should go, isn't it?
https://t.co/wdWfZDowgj
A regulator that stopped regulating <the @ICOnews
‘The problems didn’t begin with John Edwards’: Pressure grows for wider overhaul of the ICO <it’s been due an overhaul since at least Liz Denham’s tenure https://t.co/l3ypDcoBs7
The intentional normalisation and scope creep of digital #surveillance of learners’ behaviour and activity in #education, into the monitoring of employees in the workplace, is now open for consultation — https://t.co/hClLfAFdpA #AI#privacy#DataProtection Closing September 30th.
🇪🇺 The European Parliament lets #ChatControl 1.0 go through despite a majority voting against it (314:276) – allowing mass scanning of private chats until 2028. Survivors are warning. My analysis and why this is the wrong approach 👇
https://t.co/kcvgTfW8HY
#DemocracyFail
Companies will once again be allowed to scan citizens’ personal texts, emails, and social media messages to find child abuse material online. https://t.co/0iJFvsAvLO
The Government has confirmed that its so called 'independent' Digital ID Advisory watchdog, will not have its minutes published.
We are expected to carry out parliamentary scrutiny on one of the most significant erosions of our civil liberties for decades, and yet we are not told the group budget, member selection process, or given the minutes from their meetings. Journalists have also been excluded from advisory panels.
The Government must give us answers on this sinister policy, and rethink its current course.
https://t.co/lFNq3lcyeV