Yesterday I wrote, “Pressure is building for extra power stations to help out AI techbros, and with the cost put on consumers’ bills.”
However Anthropic, the developers of Claude, have pledged to “pay 100% for grid upgrades needed to interconnect our data centres… and this includes any costs that would otherwise be passed onto consumers.”
They would do this by paying a monthly surcharge themselves to the electricity and grid supply companies.
This does not appear to have gone down very well with the techbro-friendly Trump administration!
Worse than that, according to Axios, a US news website, based in Arlington, Virginia (a 15 minute drive from CIA HQ) a more serious problem has arisen.
Anthropic appear to be the only AI supplier at the moment able to supply the US armed forces with security cleared approved products. Negotiations between the Pentagon and Anthropic on signing the supply deal have not yet been finalised.
Anthropic wish the use of their products by the Pentagon to be subject to two restrictions. Although Anthropic is prepared to relax its current terms of use to enable military use, it wants (1) to ensure its products aren’t used for unrestricted surveillance of Americans, and (2) nor to develop autonomous weapons that fire to kill without human involvement.
Such automatic robokillers could use facial recognition software to identify their targets – for example.
Peter Hegsmith, Secretary for War, is very upset indeed and claims that any such restrictions are unacceptable to the Pentagon and threatens to designate Anthropic as “a supply chain risk”. This would prevent them dealing with the US government and all US government suppliers would have to cease dealing with Anthropic.
Looking at the situation the other way round, it is clear Hegseth really does want to keep Americans under surveillance and/or with autonomous robot killer machines armed with AI information hunting down and eliminating human targets.
Stay safe, good people!