Video · 5 October 2026 · 2 min read

Open AI Models and the Safety Debate

Abdul Osama (عبدول أسامة) discusses a view that calls to slow the most advanced AI models may overlap with providers’ concerns about competition from open models. He suggests that businesses paying heavily for hosted AI might consider running models locally, but does not establish that commercial interests are driving the safety calls.

Key takeaways

Abdul presents commercial self-interest as a possible explanation for some safety calls, not an established fact.
He connects that possibility to competition from models businesses can run themselves.
He suggests local models as an option for companies spending heavily on hosted AI.
The video does not establish whether a local setup would cost less for a particular business.

What is the debate?

Abdul Osama (عبدول أسامة) asks whether calls to slow the development of the most advanced AI models are solely about safety or might also reflect competition from open models. He presents the commercial explanation as a view held by some people, not as a proven account of the companies’ motives.

Abdul says Dario, the head of Anthropic—an AI model developer—wrote an article urging restraint with frontier models, or the most advanced AI systems, because of a potential loss of control. He says Sam Altman and Elon Musk expressed agreement. Abdul then questions whether that alignment could also reflect concern about businesses adopting open models.

“A lot of people see it as an attempt to slow large American companies’ use of open-source models.” — Abdul Osama

Why might businesses consider local models?

In this discussion, Abdul uses “open-source models” for models an organization could run on its own infrastructure rather than relying entirely on an outside provider. He suggests that companies paying substantial amounts to OpenAI or Anthropic might instead consider buying hardware, setting up a data center or IT room, and having IT staff operate local models. He also mentions AMD, a computing-hardware company, when discussing hardware that could make local use more accessible.

| Option | How Abdul describes it | | --- | --- | | Hosted model service | Keep paying to use models from providers such as OpenAI or Anthropic. | | Locally run model | Buy equipment, set up infrastructure and have IT staff operate the models. |

“They’re all thinking about businesses.” — Abdul Osama

In numbers

Abdul mentions 48 months as a possible timeframe for losing control, but phrases it uncertainly. The video provides no evidence for the estimate.
His example of a local setup includes one or two IT specialists to operate the models.

Background

“Open” does not mean every model allows every use: licenses and hardware requirements vary. Owning equipment also brings operating and maintenance costs.

To assess the issue, separate the safety question from the business decision between a hosted service and a locally run model. The video offers an interpretation of companies’ incentives, not evidence that settles their motives.

Questions and answers

Why does Abdul question AI leaders’ agreement on safety?

He relays a view that their shared position could also protect their businesses from competition by open models. He does not establish that this is their motive.

What does it mean to run an AI model locally?

It means operating a model on infrastructure an organization owns or manages, rather than relying entirely on an outside provider’s service.

Why would a company consider local models?

Abdul suggests that businesses paying substantial amounts for hosted models might consider buying equipment and operating models themselves. He gives no cost comparison that determines which option is cheaper.

Does the video show that open models are safe or unsafe?

No. It discusses the safety debate and possible business incentives, but does not assess the safety of a specific model.

Sources and tools mentioned

OpenAI

Full transcript

احنا قدام مسرحية لذيذة قوي شركات الزكاء الصناعي التلاتة الاكبر في امريكا بيلعبوا علينا لعبة. داريو المدير الأنثروبيك كتب مقالة من تلاتة اربعة يام قال ان احنا لازم نهدي اللعبة في الفرنتير مودلز المودلز الاقوى عندنا عشان البشرية في خطر واحنا في خلال تمانية واربعين شهر او whatever ممكن نفقد السيطرة تماما وغريبة جدا سام ألطمين عدوه اللدود اللي هم كانوا شغالين مع بعض وبعدين داريو ساب وفتح أنثروبيك قال انا I agree with Darيو وبعدين دخل إيلون موسك قال وانا كمان وماشاء الله حبوا بعض خالص. مسرحية لذيذة جدا بس من وجهة نظر ناس كتير هي مجرد محاولة منهم لان هم يفرملوا استخدام الشركات الامريكية الكبيرة للopen source models الصينية او الopen source models عموما وده عشان هم حاسين ان الopen source models خطر على البيزنس بتاعهم فعايزين يعملوا قوانين في امريكا والعالم ان الشركات الكبيرة ما تستخدمش الopen source عشان هو خطر على البشرية فهتستخدم ايه بقى؟ تستخدم شركاتهم فيبقوا in business لان هم واضح ان هم ممكن بالشكل ده وبالانتشار بتاع الopen source models وتعلم الناس لاستخدامها والاجهزة اللي بتعملها AMD وغيرها من الشركات بتاعت الهاردوير اللي eventually هتخلينا نعرف نrun الlocal models على اجهزتنا طبعا حاجة consumers اخر من في حساباتهم كلهم بيفكر في الشركات الشركات دلوقتي اللي لقيت نفسها بتدفع جامد جدا مع open AI او Anthropik او X قالك طب نحنا نرجع بقى على قديمه يبقى عندنا data center او عندنا قوضة IT فيها تكييف ونجيب اجهزة ندفع مرة واحدة ونrun الlocal models وعليها واحد بتاع IT ولا اتنين بيفهموا يشغلوا الموديلز ديات وامورنا طيبة مسرحية دايرة ويخشوا على تويتر شوفوا بقى الناس بتقولهم ايه وهم ما شاء الله كلهم حبوا بعض مرة واحدة مصلحتهم اجتمعت طبعا يعني شوية اخبار كده على الصبحة انا عبدالأسماء تابعنا عشان تعرف كل جديد في الزكاء الصناعي والتسويق.

This text was written with AI from the video and may contain small transcription errors. · Markdown version

Watch next

Why Are AI Companies Moving Into Cybersecurity?

Abdul Osama (عبدول أسامة) links Anthropic, OpenAI and Cursor’s interest in cybersecurity to two problems: vulnerabilities in rapidly developed, insufficiently reviewed software and older flaws that newer AI models can uncover. He says the companies are moving toward security solutions for enterprises and people who build software with AI.

5 October 2026 · 1 min read

Running AI Models Locally: What Hardware Is Needed?

Abdul Osama (عبدول أسامة) ran an AI model on his gaming PC without internet access and got replies, but found it slow and limited. He says running a much larger open-source model yourself could require costly hardware; renting hardware to host a model is another option.

5 October 2026 · 2 min read

Are AI tools becoming more expensive?

Abdul Osama (عبدول أسامة) expects some AI tools to become more expensive and predicts fewer benefits in cheap or free plans. He suggests running open-source models on owned or rented hardware as an alternative, while stressing that it takes learning and technical work.

5 October 2026 · 2 min read