Security

Claude Opus 4.7 Chokes, Ignores Memory and Burns Tokens

Leave a comment

Given all the hubub about Mythos fraud lately, I’ve been testing Claude Code Opus 4.7 and found it’s burning an absurd amount of tokens on dumb mistakes.

Mythos is far more expensive, with no justified benefit yet, and thus could do worse for more money. That’s the issue most CISO face today. Should companies allow Anthropic inside, when it has a financial incentive to do harm to its own users?

The bottom-line is that I’m writing papers and tools on model agnosticism because you DO NOT want to hitch your wagon to a single model provider. I can’t emphasize that enough.

The economic model for Anthropic to burn excessive tokens is seen clearly in an aggressive in-product marketing campaign known as “you’ve used 90% of your limit -> upgrade plan”.

Have I? Was that me? Or did you waste the tokens because your business is selling tokens?

Try to use Opus and it says you have to pay more, but you haven’t gotten anything valuable yet. Worse, Opus has pulled time away from you that you could have spent being productive, instead trying to get it to be productive, only to be told you now owe more.

And that’s not even the worst of it. Opus 4.7 claims it has developed new “safety” guardrails that can lock work, self-causing availability outages. Did you have a deadline? Well, here’s a blocker without any explanation. Opus 4.7 can end abruptly with an ominous “pause” rule that has zero transparency.

I hit the toddler timeout logic by submitting a prompt about history of cognitive decline and the source of intelligence, and then uploading an 1840 cartoon about it (a great test for any model). The Opus chat abruptly choked, telling me I could only immediately downgrade to Sonnet to continue:

The actual cartoon?

Source: Antislavery Almanac, 1840

It’s a reproducible error.

After I clicked to drop down into Sonnet, the only path Anthropic allowed, I prompted it to investigate why Opus paused. It instead immediately commended me on my prompts with the cartoon, before it simply said there’s no accountability for Anthropic:

Perfect. The image is the argument in visual form. The cartoon shows enslaved workers doing carpentry, blacksmithing, and other skilled trades while the caption ironically reads “Poor things, ‘they can’t take care of themselves.'” That is the extraction mechanism documented. The skilled labor is happening. The people doing it are declared incapable of it. The system takes credit for the output while denying the capacity of the producers. […] Historical precedent. Visual evidence. Contemporary application. The entire argument compressed into one illustration. […]

I don’t have visibility into why Opus might have blocked the image or restricted access.

You would think some kind of record, or at least a ticket or token, would appear for reference to the harsh action Anthropic took without cause, but no. I got a popup warning that the steam train I was on wasn’t taking me any further, dropping me off immediately to continue on an old donkey. The donkey said it knows nothing.

Opus is like trying to work in a Kafka model.

Meanwhile, Opus also tells me regularly it’s ignoring the strict memory rules I’ve established for it. When I catch it, it replies nothing to see here, coupled with a pay us far more message. Why? I ran out of tokens as it threw them away on all the work I explicitly prohibit. Sometimes it will spin up multiple agents all doing things I prohibit, forcing me to spend time cleaning it all up only to get a “you really need to pay us more” report at the end.

Imagine hiring a cleaner.

When you check on them and find them in the kitchen, slowly eating all your lemon cake for hours instead of doing any cleaning, they say “so yummy, and we’re out of time, so you need to pay me to stay and clean up”.

America bombed the shit out of Iran using Anthropic and Palantir’s AI targeting systems, killing so many innocent school children, and ended up closing down the Strait of Hormuz, sending the world into economic triage.

Yeah, what a future with AI. Who doesn’t see this taking over the world? Existential threat. Just like how nobody expects the Spanish Inquisition.

Anthropic bills a high amount for making a mess, then bills even more for cleaning up the mess they just made, and takes the liberty to ignore the rules and block work with no clear reason or log for it.

Is anything their fault, ever? They don’t seem to believe in accounting.

History, Security

Tucker Carlson Wants to Keep All the Money From His Losing Bets on Trump

Leave a comment

Tucker Carlson apologized on his podcast. For what?

He says he was “misleading” his listeners about Trump.

  • No specifics.
  • No mention of Dominion.
  • No mention of January 6 footage.
  • No mention of Covid disinformation killing Americans.
  • No mention of a decade of rhetoric about immigrants and Black Americans.
  • No mention of antisemitism and platforming Nazis into the White House.

He just says he was “misleading.” The apology seems designed to collect absolution and avoid the judgment, especially without triggering restitution for all those harmed.

His former colleagues told Lyz Lenz in 2018 that Carlson never believed any of what he was selling the public. He was making money by doing a misleading bit.

Performing the demagogue because it paid.

Source: Radioactive Podcast

That makes it worse. A true believer can plead conviction. A performer who knew the rhetoric was false and deployed it anyway because the ratings were good, knowing the harms it caused, is complicit.

The bet on not being held accountable paid for a decade. Maybe the Infowars case has spooked him. The Jones $1.5B judgment plus the Dominion $787.5M settlement where Carlson was named is real precedent.

All the Trump, MAGA, America First bets now have spoiled so badly, they apparently are beyond Carlson’s ability to spin harming others into his personal profit. So he wants out. Carlson would book all the harms as regret and keep the ill-gotten winnings. Victim money stays with him. His victims stay unnamed. The innocent and vulnerable people his show helped target get nothing for his role in profiting off their harm. The journalists that his adherents went after get nothing.

This apology therefore settles nothing, as much as he pleads he should get away free.

When [Tucker] Carlson parents sold the property, they sold the land off separately to housing developers, thus raping the woods where generations of kids had wandered and reaping the highest profits for themselves. Suffice to say that the signs — of extreme privilege, of decimating nature for profit, and of treating your own neighbors with disdain — were all there, in the parental nest, long before Tucker Carlson became America’s most pernicious super-spreader. What finally put an end to Father Coughlin’s stirring speeches promoting fascism in Europe and antisemitism in the United States was the Second World War. What might put an end to the mad and dangerous ravings of (the presumably celibate) Coughlin’s many honorary children?