[24155] in Privacy_Forum
[ PRIVACY Forum ] Script of my national radio report yesterday
daemon@ATHENA.MIT.EDU (Lauren Weinstein)
Tue Aug 4 10:31:57 2026
Date: Tue, 4 Aug 2026 07:22:33 -0700
From: Lauren Weinstein <lauren@vortex.com>
To: privacy-dist@vortex.com
Message-ID: <20260804142233.GA11854@vortex.com>
Content-Disposition: inline
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset="us-ascii"; Format="flowed"
Errors-To: privacy-bounces+privacy-forum=mit.edu@vortex.com
This is the script of my national network radio report yesterday
discussing new instances of AIs attacking AIs, and more abuses by AI
firms. As always there may have been minor wording variations from
this script as I presented the report live on air.
- - -
Yeah, recently I discussed how OpenAI's AI model attacked another AI
firm, not due to any flaw with the AI, but because OpenAI had sloppily
not set their testing conditions to properly limit the AI's actions.
Now we hear that another AI firm, Anthropic, found their Claude AI
doing the same sort of thing, again due to corporate sloppiness in
setting controls.
A person could go to prison for years doing attacks like these. The AI
firms get away with it, as usual, since they know how to play the
political games. They want us to trust their AIs monitoring our
meetings. They assure us that they won't leak proprietary or personal
information. They want us to let their AI agents run amok on work and
personal computers, not only reading data but actually taking actions
on systems and web sites on behalf of users. But these recent
incidents show that there's no reason to believe that doing any of
this is actually safe, especially since the firms routinely refuse to
take any responsibility for anything that goes wrong. And when firms
can't even get their own internal AI security controls to work
properly, what possible reason would there be to trust them regarding
any of this stuff?
Here's another new one. Google used to have a world class privacy team
and a top-notch trust and safety team. But maybe most of the team
members were laid off and replaced with AI, because Google really
screwed up last week. Google announced that anyone could now create
fake imagery using Google AI and combine it with Google Earth imagery.
Google Earth has long been the gold standard for reliable geographic
images. But immediately horrific deepfake images started to circulate
with people using this new AI Slop feature to create realistic imagery
portraying disasters and other horrible and misleading situations.
Google's initial reaction to concerns about this dangerous
misinformation was to try downplay the problem, but as warnings and
examples built up Google relented and ultimately pulled down the
feature entirely -- it was running less than 24 hours. Yet Google has
made it clear that they want to bring it back, despite the entire idea
being incredibly dangerous, no matter what new so-called guardrails
they might try to implement.
One really has to ask, how did such an obviously disastrous Google
Earth misinformation AI Slop feature get green-lighted in the first
place? And by the way, there's new reports now that the AI industry
has been quietly obtaining enormous numbers of physical books --
sometimes rare ones -- that they then destroy in the process of
scanning them into their AIs. Another way to thumb their noses at
society.
It's clear that these firms cannot be trusted to safely manage these
AI technologies by themselves. They need to be firmly regulated with
an eye towards society's needs, not the bottom lines of the Big Tech
billionaire CEOs pushing these technologies. These firms have
benefited from a lot of slack up to now thanks to their political
connections with both parties, but it's way past time to tell these
firms that their wild partying days are over, and that they're either
going to play by society's rules, or they're not going to be permitted
to play at all.
- - -
L
- - -
--Lauren--
Lauren Weinstein
lauren@vortex.com (https://www.vortex.com/lauren)
Lauren's Blog: https://lauren.vortex.com
Mastodon: https://mastodon.laurenweinstein.org/@lauren
Signal: By request on need to know basis
Founder: Network Neutrality Squad: https://www.nnsquad.org
PRIVACY Forum: https://www.vortex.com/privacy-info
Co-Founder: People For Internet Responsibility
_______________________________________________
privacy mailing list
https://lists.vortex.com/mailman/listinfo/privacy
