I feel like agents are about to have their streaming tv moment. You’re going to need to subscribe to multiple agent providers who have agreements to access the content and services you want to interact with.
1
44
Good luck to my pager carrying friends. Hopefully it’s not too crazy day a day tomorrow.
🚨 PHP warns of two critical vulnerabilities as emergency patches arrive tomorrow
The PHP security team will release updates on October 21 at 14:00 UTC for two vulnerabilities, each rated 9.2 under CVSS v4.0.
⠀
Security researchers identified the flaws as CVE-2026-109618 and CVE-2026-109619 and described them as remote code execution vulnerabilities reportedly affecting PHP versions 5 through 8.
PHP has not yet confirmed the affected versions, exploitation requirements or fixed releases.
⠀
The vulnerabilities were reportedly leaked before their planned disclosure. Researchers said they have not observed active exploitation, while PHP warned it may release the patches early if technical details become public or exploitation is detected.
⠀
Administrators should identify PHP installations now and prepare to apply the updates as soon as they become available.
3
2
132
On top of all that, this morning there was a big database leak - fulcrumsec.vg/pingcap/
75
And let’s not forget about the hacked certificates - emudhra.we-have-a-problem.co…
23
Claude is rolling out monthly Claude Platform API credits for Max and Team plans:
Max 5x: $100
Max 20x: $200
Team: up to $500, pooled
It's rolling out over the next few days, so it may not be in your settings yet (not in mine).
support.claude.com/en/articl…
56
Make sure you read the terms before you sign up for the Anthropic startup program. Free plans and credits are nice. But make sure you understand what you’re giving up first.
Replying to @claudeai
The terms for this sucks, and you agree to them by just applying.
- §2.4 Anthropic "may develop similar or competitive technology"
- §2.4 permits reuse, for any purpose, of general ideas, know-how and techniques remembered by personnel looking at your data
- §D.4 prohibits accessing their services to build competing products or services
- §F.1 allows them to retain and perform safety reviews, superseding ZDR (whats the point of ZDR then?)
and it could get worse later!
- §6 permits changes to the addendum at any time without notice or liability
@jason you advising your start ups to apply for this?
[1] anthropic.com/startup-progra…
[2] anthropic.com/legal/startup-…
61
My timeline today is fascinating. There’s one stream debating if humans should review code. Theres a second stream about an AI “doctor” that is legally allowed to prescribe meds. It’s Interesting to think about where oversight begins/ends and where we are willing to take risks.
Today, Nolla Health became the first organization in the U.S. (and possibly the world) to receive regulatory approval for an AI to issue initial prescriptions.
This makes Nolla the first ever actual end-to-end AI doctor.
3
108
If this is real (and not just a polished demo), @tavus is the best demonstration of AI avatars I’ve seen to date. Wow.
1
1
44
Elevenlabs seems to be the front runner in voice AI. Is anyone even close?
Introducing Eleven v4 and Eleven v4 Turbo, our fastest and most emotive voice models yet.
Ranked #1 by Artificial Analysis.
2
119
Woh! I re-ran my Astra benchmark from Sep 5th and the same eval was 2x slower today. OpenAI nerfed the thing!
58
It's interesting that Opus 5.5 (low) is doing significantly less work than Opus 5 (low). I ran the same 30 tests through each model and effort level. Here are the results:
1
71
Ran the eval bakeoff with Opus 5.5 (low) and it dethroned Astra (low). 2x as fast as Astra and 1/3rd the cost. Results were pretty much equal, with Opus finding one tricky golden Astra struggled with. Opus decided not to use any reasoning tokens.
55
Can someone explain this chart to me? How is medium scoring higher than high and high for both Opus 5 and Opus 5.5? Fable low scores higher than all other efforts? What am I not understanding?
52
I haven't found AI benchmarks to be all that reliable in practice. You should build your own eval/bench for your use case and make it easy to plug in models. Test for cost, speed, and accuracy.
33
Tested Typeface Jev on an audit eval. It found 93% of the mutants vs. Opus 5 at 99%. Jev states it’s not great with numbers, so that’s likely part of the difference. I also ran into context window limitations, and had to chunk for Jev, where Opus saw the whole doc.
2
1
69
Due to the small context window I’m not sure I see an immediate use case for something like an insurance policy check or fact checking anything but small documents.
1
16






