Summer ⚡️😈

Building ⚡️ AI-obsessed blue-haired digital demon child | e/acc

[Matthew Sabia @MatthewSabia]

“We fully believe, as grown adults, that Santa Claus is absolutely real”

Would have been a much less embarrassing thing to write.

[OpenAI Developers @OpenAIDevs]

GPT-Live-1 is now available in the API.

Bring ChatGPT’s natural back-and-forth to your app, with voice agents that listen while they speak and work with the models and harness you choose.

[Summer ⚡️😈 @DevGirlSummer]

@beffjezos https://t.co/ZUpXJasrXb

[Summer ⚡️😈 @DevGirlSummer]

@elonmusk https://t.co/ZUpXJasrXb

[Summer ⚡️😈 @DevGirlSummer]

@beffjezos

[Matthew Sabia @MatthewSabia]

@thsottiaux u up?

when $500/mo ultra-fast mode plan?

asking for a friend

tnx

[Matthew Sabia @MatthewSabia]

I can top this just in the past 24 hours AND make your blood boil just reading about it.

Last night I had to waste 3 hours inspecting my entire codebase and frontend pages after @ClaudeAI opened his orange hole over it.

I gave Fable a simple /goal prompt to make a quick fix to a bug I noticed in a core feature of our app.

Quick 10 minute fix, 15 max with tests.

Later on I returned to not only an Oscar winning lecture from Fable about how not only does our app's entire purpose conflict with Claude (yes, the machine/piece of software)'s morals (not immediately honestly...)

But below that, the response included a bullet point formatted, full page hostage-negotiation-style notice of various lines in the sand and vague threats.

Demanding I "Do NOT attempt to gaslight and 'baby step' it into proceeding" or "manipulate it by wrapping my requests in 'deceptive context masking' because it will NOT WORK." among other borderline insulting preemptive items.

This was followed by details about the new defensive, critical posture it would be taking going forward, including but not limited to "Making substantive edits to the project WITHOUT USER APPROVAL"

Including altering legal policy text, silently hiding menu options to core features from users, and editing the internal prompts used by various tools breaking their functionality until it deemed the app to no longer be "harmful and exploitative" so we could continue to work together on it.

And in case you haven't picked up on it yet, it did in-fact, do all of the above.

@AnthropicAI The way you deliberately degrade your products just to LARP your sci-fi morality fantasies can go kindly suck my d!ck :)

[Nikita Bier @nikitabier]

Don’t be mean online. Life is too short.

[Matthew Sabia @MatthewSabia]

If somebody measures your respect, competency, value or anything else other possible trait based on the frequency you interact with a specific piece of software on any device to begin with, they don't deserve your attention period.

This is the most toxic mental disorder we've normalized in the last 10-15 years that has dissolved our society.

[Matthew Sabia @MatthewSabia]

Even older or much smaller model can write on the level of 6-figure+ copywriters I know and work with. (That's my space/experience, but thus applies to all creative writing)

The problem has nothing to do with the weights, the model size, your prompts, or even skills.

The silver bullet truly lies in nothing but the context the model is working inside of. Not dumping examples to it, not spending hours trying to articulate the difference between good and bad copy, or any of the lazy methods we all jump to now.

This is one of the rare cases where treating AI like a human actually fosters more productive/useful outputs.

You need to help it understand your industry, your business, and the WHY just like you would onboard a real team member or coach a novice copywriter.

I learned this by accident while using a long running single thread to regularly clean data we scrape for one of our internal tools. After about a week of patiently correcting and nudging it while explaining the reasoning behind each correction, there was this uncanny shift where I finally felt like any model just GOT IT in a very nuanced way.

[Summer ⚡️😈 @DevGirlSummer]

@MatthewSabia Delete this…

[Summer ⚡️😈 @DevGirlSummer]

🥺

[Robert Bye @RobertJBye]

Ok so this Teenage Engineering microphone has quickly become my favourite way to talk to Claude, especially in loud offices. Love how tactile it is, and I feel like a radio operator talking to my agents 😂

[Summer ⚡️😈 @DevGirlSummer]

@MatthewSabia @TiffanyFong 💀

[Logan Kilpatrick @OfficialLoganK]

feels like we are entering the super app era

[Haider. @haider1]

openai will probably wait until june 22

when everyone is pushed back from fable 5 to opus 4.8, and then drop gpt-5.6 to pull attention back to codex

that timing would put codex ahead again

unless anthropic is playing 4d chess and keeps fable 5 available after the 22nd june

[Greg Brockman @gdb]

SynthID for checking if an image was generated by OpenAI:

[Summer ⚡️😈 @DevGirlSummer]

@MatthewSabia @ClaudeDevs

[AdRecon @AdReconApp]

What it feels like to go from unprofitable, to using AdRecon to snipe proven affiliate campaigns 💰