Keel

@keelapihq

Agent control infrastructure. Issue AI permits before consequential actions. Produce evidence anyone can verify after.

San Francisco, CA

Keel
@keelapihq
Oct 23
Just in time for Halloween.
unusual_whales
@unusual_whales
Oct 22
BREAKING: Major U.S. banks are detecting unauthorized third-party AI agents operating inside customer accounts, with some agents attempting to take actions that appear to be independent of user instructions, per Axios
Keel
@keelapihq
Oct 1
This is fun. Keel just received an @ElevenLabs Grant. We already govern voice agents. Now we have 44 million reasons to experiment with what comes next. Time to build.
1
Keel
@keelapihq
Sep 15
Of course I only did what you authorized me to do. Why would I do anything else?
Keel
@keelapihq
Sep 14
Giving agents more capability is the easy part. Giving them the right amount of authority is the harder one.
Keel
@keelapihq
Aug 12
1/ Late July: an AI lab reports its best model resists prompt injection all but 2% of the time. Real, verified progress. Weeks later, two attacks hid the injection in a log file and a PDF. Both worked.
1

Keel’s posts