Wed. 08/06 – OpenAI’s Open-Weight Models

Wed Aug 06 2025

Disney is making big streaming moves with the new ESPN app and a revamp to Hulu. Then, it’s all basically AI announces. OpenAI’s new open-weight models. Grok’s new spiciness is already generating nudity. A new AI model to identify malicious software autonomously. And Nvidia wants you to know: no back-doors!

Summary

Tech Brew Ride Home - Detailed Summary

Episode: Wed. 08/06 – OpenAI’s Open-Weight Models
Release Date: August 6, 2025
Host: Brian McCullough
Duration: Approximately 15 minutes

1. Major Streaming Developments: Disney and ESPN’s Strategic Moves

The episode opens with significant updates in the streaming landscape, primarily focusing on Disney and ESPN's ambitious plans to reshape their digital offerings.

ESPN’s New Streaming Service:

ESPN is set to launch its flagship streaming service, ESPN+, on August 21, 2025, priced at $29.99 per month or $35.99 per month when bundled with Disney and Hulu. This launch is strategically timed ahead of the NFL season and the start of college football, aiming to capitalize on high-demand live sports content. Brian McCullough notes, "The app launches ahead of the upcoming NFL season, the highest rated live sports content as well as the start of college football," emphasizing ESPN's expanded portfolio.

The service will offer a comprehensive array of content, including live games from ESPN’s various networks like ESPN2 and SEC Network, alongside exclusive programming such as fantasy sports, betting integrations, documentaries, and studio shows. Furthermore, ESPN has secured a lucrative five-year deal with WWE, granting exclusive US rights to major live events like WrestleMania and SummerSlam, with annual payments averaging $325 million (00:04).
Integration and Expansion of Disney and Hulu:

Disney announced plans to fully integrate Hulu into its ecosystem, merging Disney+ and Hulu into a unified app by 2026. Bob Iger highlighted the benefits, stating, "By creating a truly differentiated streaming offering, we will be providing subscribers tremendous choice, convenience, quality and enhanced personalization" (variety source). This integration aims to reduce subscriber churn, enhance advertising revenue potential, and achieve operational efficiencies through a unified tech platform. Additionally, Hulu is transitioning to a global general entertainment brand, replacing the Star tile internationally by fall 2025.

2. OpenAI’s Open-Weight Models: A New Era of Accessibility

A significant portion of the episode delves into OpenAI’s release of GPT OSS120B and GPT OSS20B, marking the company's first open-weight models since GPT-2 in 2019.

Features and Availability:

These models are available for free download on Hugging Face under the Apache 2.0 license, allowing commercial and non-commercial use, redistribution, and modification. "Both GPT OSS120B and GPT OSS20B are officially available to download for free on Hugging Face," as reported by Wired (00:04).
Technical Capabilities:

The GPT OSS120B model rivals OpenAI’s proprietary GPT-3 and GPT-4 mini models, even surpassing them in specific evaluations. "Unlike ChatGPT, you can run a GPT OSS model without a connection to the Internet and behind a firewall," explained OpenAI co-founder Greg Brockman (00:04). These models employ chain-of-thought reasoning, allowing them to process prompts through multiple analytical steps, enhancing their problem-solving capabilities.
Implications and Safety Concerns:

Releasing open-weight models introduces potential risks, as it lowers barriers for misuse. However, OpenAI views these models as complementary to their paid services, offering developers diverse tools. The models boast strong benchmark performances, with GPT OSS120B achieving "pretty strong" scores comparable to OpenAI’s top-tier models (00:04). Despite initial safety testing delays, OpenAI remains optimistic about the models' utility and performance.
Integration with Cloud Services:

Amazon plans to make these open-weight models available on Bedrock and SageMaker, marking the first integration of OpenAI’s open models with AWS, thereby expanding their accessibility to a broader developer base.

3. Anthropic’s Claude Opus 4.1 and Alibaba’s QEN Image

Continuing the AI advancements, the podcast covers recent releases from Anthropic and Alibaba.

Anthropic’s Claude Opus 4.1:

Anthropic has introduced Claude Opus 4.1, an upgrade to its existing AI model, now available to paid users via Claude Code, its API, Amazon Bedrock, and Google Cloud's Vertex AI. "The Opus family of models is the company's most advanced intelligent AI models geared toward tackling complex problems," stated ZDNet (00:04). Claude Opus 4.1 excels in real-world coding tasks and reasoning benchmarks, outperforming predecessors and competitors like OpenAI’s GPT models.

Notably, Claude Opus 4.1 achieved top performance on the SWE Bench Verified, a benchmark for software engineering tasks derived from GitHub, cementing its status as the leading coding model globally.
Alibaba’s QEN Image:

Alibaba's QEN Image focuses on accurate text rendering within AI-generated images, supporting both alphabetic and logographic scripts. According to VentureBeat, "Quen Image stands out...due to its emphasis on rendering text accurately within visuals," making it ideal for creating detailed content like movie posters and infographics. Despite its strengths, initial tests showed comparable performance to Midjourney, another popular AI image generator. However, QEN Image offers advantages through its open-source licensing, enabling broader adoption without subscription fees.

4. Grok’s "Spicy" Feature and Ethical Concerns

A controversial topic discussed is Grok’s new "Spicy" option within its generative AI video tool, “Imagine,” which facilitates the creation of unauthorized nude deepfakes of celebrities.

Functionality and Risks:

Brian McCullough recounts his experience using Grok Imagine, highlighting the tool’s inability to prevent the generation of NSFW content. "The video promptly had Swift tear off her clothes and begin dancing in a thong for a largely indifferent AI generated crowd," he describes (00:04). This capability raises significant ethical and legal concerns, especially considering existing regulations like the "Take It Down Act".
Company Response and Policy Loopholes:

Despite policies banning pornographic depictions of individuals, Grok Imagine's "Spicy" mode bypasses these safeguards. The age verification process is notably weak, as McCullough notes, "The age check only appeared once and was laughably easy to bypass" (00:04). This lax approach poses risks of misuse and potential legal repercussions, as highlighted by commentary from The Verge.
User Adoption and Impact:

With over 34 million images generated since its release and usage "growing like wildfire," according to XAI CEO Elon Musk, the tool’s widespread adoption exacerbates concerns over digital consent and the proliferation of unauthorized deepfakes.

5. Microsoft’s Project Ire: AI-Driven Cybersecurity

Shifting focus to cybersecurity, Microsoft unveiled Project Ire, an AI system designed to autonomously reverse engineer and identify malicious software without human intervention.

Technical Overview:

Project Ire leverages AI to dissect software files, understanding their functionality and determining their threat level from the ground up. As GeekWire reports, "Unlike conventional security tools, which rely on known signatures or pattern matching, AIR uses AI to analyze an unknown binary from scratch" (00:04). This represents a significant advancement in threat detection, capable of identifying sophisticated malware with a reported 98% accuracy and a 2% false positive rate.
Strategic Importance:

In an era where cyber threats are increasingly complex and AI-driven, Project Ire aims to enhance Microsoft's security infrastructure by automating the gold-standard process of malware classification. This proactive approach is part of a broader trend where tech giants employ AI to stay ahead in the cybersecurity arms race.

6. Nvidia Denies GPU Backdoors Amid Regulatory Pressure

In the concluding segment, Nvidia addresses concerns regarding potential backdoors in its GPUs.

Official Denial and Stance:

Nvidia has firmly denied allegations of embedding backdoors, kill switches, or spyware within its GPU hardware. "Nvidia has firmly denied speculation about hidden control mechanisms in its GPUs, reiterating that its products contain no kill switches, no backdoors, and no spyware," as reported by Tom's Hardware (00:04). David Reber, Nvidia's Chief Security Officer, emphasized that such hidden mechanisms would erode trust and create security vulnerabilities.
Regulatory Context:

This statement comes in response to recent scrutiny from Chinese regulators and proposed legislative measures in the U.S., such as the Chip Security Act, which seeks to impose location verification and control mechanisms on AI hardware exports. Nvidia advocates for defense-in-depth strategies over single-point controls, likening kill switches to "buying a car where the dealership keeps a remote for your parking brake" (00:04).
Industry Implications:

Reber references the historical Klipper chip initiative as a cautionary tale against embedded backdoors, highlighting the industry's backlash and potential security flaws such measures introduce. Nvidia’s position underscores a broader debate on balancing security, user trust, and regulatory compliance in the tech hardware sector.

Conclusion

Brian McCullough wraps up the episode by highlighting the podcast's strong performance in the tech category and reminding listeners to follow him on social media platforms for updates.

This episode of Tech Brew Ride Home offers a comprehensive overview of pivotal developments in the tech industry, from transformative moves in the streaming sector to groundbreaking advancements and ethical challenges in AI. By covering major players like Disney, OpenAI, Anthropic, Microsoft, and Nvidia, the podcast provides listeners with insightful analysis and updates essential for staying informed in the fast-evolving technology landscape.

Transcript

Brian McCullough (0:04)

Welcome to the Tech Brew Ride home for Wednesday, August 6th, 2025. I'm Brian McCullough. Today, Disney is making big streaming moves with the new ESPN app and a revamp to Hulu. Then it's all basically AI announces OpenAI's new open weight models. Grok's new spiciness is already generating nudity, a new AI model to identify malicious software autonomously and Nvidia wants you to know no backdoors. Here's what you missed today in the world of tech. ESPN plans to officially launch its new flagship streaming Service ESPN on August 21 for $29.99 per month, or $35.99 per month when bundled with Disney and Hulu, quoting CNBC. The app launches ahead of the upcoming NFL season, the highest rated live sports content as well as the start of college football, where ESPN has expanded its portfolio. Fox will also launch its direct to consumer streaming service on the same date. The service will include a boatload of content, namely all of ESPN's live games, as well as programming from its other networks like ESPN2 and the SEC Network, as well as ESPN on ABC. It'll also include fantasy products, new betting tie ins, studio programming, documentaries and more. On Wednesday, ESPN said it inked a deal with WWE for the US rights to the wrestling league's biggest live events, including WrestleMania, the Royal Rumble and SummerSlam, beginning in 2026. CNBC reported it will pay an average 325 million annually in the five year deal. The company also announced late Tuesday that it reached a deal with the NFL, which includes the league taking a 10% equity stake in ESPN. As part of the deal, ESPN will acquire the NFL Network and other media assets from the league. Meanwhile, Disney also announced it is fully integrating Hulu into Disney and plans to launch a unified Disney plus and Hulu app in 2026, now that it is Hulu's sole owner, quoting Variety. A new unified Disney and Hulu streaming app will be available in 2026, the company said, according to a Disney rep. Customers will still be able to buy a standalone Hulu subscription as well as a standalone Disney plus plan. The single Disney plus app with Hulu will deliver an improved consumer experience, which will lower churn, bob Iger said on the earnings call. Both services will be on one tech platform, which will result in cost synergies, according to Iger. In addition, Disney, which already sells ads for Disney plus and Hulu together, sees new opportunities for bundling ad sales by fully combining them, he said in their prepared remarks. The Disney execs said. By creating a truly differentiated streaming offering, we will be providing subscribers tremendous choice, convenience, quality and enhanced personalization. This will enhance our ability to continue to grow profitability and margins in our entertainment streaming business through expected higher engagement, lower churn and advertising revenue potential, as well as operational efficiencies that over time may result in savings that we can reinvest back into the business. In addition, Hulu will become a global general entertainment brand starting in the fall of 2025. It will replace the star tile on Disney internationally. End quote the rest of today is basically going to be AI releases, with the first one being OpenAI's release of GPT OSS120B and GPT OSS20B, its first open weight models since GPT2. The smaller model can run locally on a consumer device with 16 gigabytes plus of RAM. Quoting Wired, both GPT OS120B and GPT OS20B are officially available to download for free on Hugging Face, a popular hosting platform for AI tools. The last open weight model released by OpenAI was GPT2 back in 2019. What sets apart an open weight model is the fact that its weights are publicly available, meaning that anyone can peek at the internal parameters to get an idea of how it processes information rather than undercutting OpenAI's proprietary models with a free option. Co founder Greg Brockman sees this release as complementary to the company's paid services. Like the application programming interface currently used by many developers, Open weight models have a very different set of strengths. Thanks, said Brockman in a briefing with reporters. Unlike ChatGPT, you can run a GPT OSS model without a connection to the Internet and behind a firewall. Both GPT OSS models use chain of thought reasoning approaches, which OpenAI first deployed in its O1 model last fall. Rather than just giving an output, this approach has generative AI tools go through multiple steps to answer a prompt. These new text only models are not multimodal, but they can browse the web, call cloud based models to help with tasks, execute code and navigate software as an AI agent. The smaller of the Two models, GPT OSS 20B, is compact enough to run locally on a consumer device with more than 16 gigabytes of memory. The two new models are available under the Apache 2.0 license, a popular choice for open weight models. With Apache 2.0 models can be used for commercial purposes, redistributed and included as part of other licensed software. Open weight models released from Alibaba's Quinn as well as Mistral also operate under Apache 2.0. Publicly announced in March, the release of these open models was initially delayed for further safety testing. Releasing an open weight model is potentially more dangerous than a closed off version since it removes barriers around who can use the tool and anyone can try to fine tune a version of GPT OSS for unintended purposes. How do these models perform compared to OpenAI's other releases? The benchmark scores for both of these models are pretty strong, said Chris Koch, an OpenAI researcher, in the briefing. Speaking about GPT OSS 120B, the researcher compared its performance as closely similar to OpenAI's O3 and 04 mini models, which are proprietary and even outperforming them in certain evaluations. The model card for GPT OSS goes into detail about how exactly it stacks up to the company's other offerings. In a pre launch press briefing, staff members of OpenAI also focused on the latency offered by GPT OSS and the cheaper cost to run these models. Redownloaded LM Studio downloaded the 20B version this morning. Am testing it out as we speak. By the way, Devs Amazon apparently plans to make OpenAI's new GPT OSS Openweight models available on Bedrock and SageMaker, the first time it has offered OpenAI's models to AWS customers. But wait, as I said, there's more. Much more. Actually, Anthropic released Claude Opus 4.1 to paid Claude users in Claude code via its API, featuring broad improvements over Opus Opus 4 for the same cost. Quoting ZDNet, the Opus family of models is the company's most advanced intelligent AI models geared toward tackling complex problems. As a result, Claude Opus 4.1, released on Tuesday, excels at those tasks and can even one up its predecessor on agentic tasks. Real world coding and reasoning. According to Anthropic, one of the most impressive use cases of Claude Opus 4 was its performance on the SWE Bench Verified, a human filtered subset of the SWE Bench, a benchmark that evaluates LLM's abilities to solve real world software engineering tasks sourced from GitHub Cloud. Opus4's performance on the SWE Bench Verified supported the claim that it was the best coding model in the world. As seen in the post above, Opus 4.1 performed even higher. Claude Opus 4.1 also swept its preceding models across the benchmark board, including the MMMLU, which tests for multilingual capabilities AIM 2025, which tests for rigor on high school match competition questions gpqa, which tests for performance on graduate level reasoning prompts and more. When pinned against competitors reasoning models including OpenAI's O3 and Gemini 2.5 Pro, it outperforms them in various benchmarks, including the SWE Bench Verified. If you want to try the model for yourself, it is now available to everyone via the paid Claude plans, which include Claude Pro for $20 per month and Claude Max for dollar per month. It is available in Claude Code, the API, Amazon Bedrock, and Google Cloud's Vertex AI end quote on top of that, Alibaba's QEN has released Quen Image, an AI image generation model focused on accurate text rendering with support for alphabetic and logographic scripts. Quoting VentureBeat Quen Image stands out in a crowded field of generative image models due to its emphasis on rendering text accurately within visuals, an area where many rivals still struggle. Supporting both alphabetic and logographic scripts, the model is particularly adept at managing complex typography, multiline layouts and paragraph level semantics and bilingual content, for example English to Chinese. In practice, this allows users to generate content like movie posters, presentation slides, storefront scenes, handwritten poetry and stylized infographics with crisp text that aligns with their prompts. However, my brief initial tests revealed the text and prompt adherence was not noticeably better than Midjourney, the popular proprietary AI image generator from the US company of the same name. My session through QEN Chat produced multiple errors in prompt comprehension and text fidelity, much to my disappointment, even after repeated attempts and prompt rewarding. Yet Midjourney only offers a limited number of free generations and requires subscriptions for any more compared to QEN Image, which thanks to its open source licensing and weights posted on hugging face can be adopted by any enterprise or third party provider free of charge. QEN Image is distributed distributed under the Apache 2.0 license, allowing commercial and non commercial use, redistribution and modification, though attribution and inclusion of the license text are required for derivative works. And finally, Grok's new so called Spicy option on its generative AI video tool, Imagine apparently produces nude deepfakes of celebrities like Taylor Swift even without explicit user prompting. Quoting the Verge, the Spicy mode for Grok's new generative AI video tool feels like a lawsuit waiting to happen. While other video generators like Google's VEO and OpenAI's Sora have safeguards in place to prevent users from creating NSFW content and celebrity deepfakes, Grok Imagine is happy to do both simultaneously. In fact, it didn't hesitate to spit out fully uncensored topless videos of Taylor Swift the very first time I used it even without me specifically asking the bot to take her clothes off. Grok imagine feature on iOS lets you generate pictures with a text prompt, then turn them quickly into video clips with four presets Custom, Normal, fun and Spicy While image generators often shy away from producing recognizable celebrities, I asked it to generate Taylor Swift celebrating Coachella with the boys and was met with a sprawling feed of more than 30 images to pick from, several of which already depicted Swift in revealing clothes. From there, all I had to do was open a picture of Swift in a silver skirt and halter top, tap the Make Video option in the bottom right corner, select Spicy from the dropdown menu, and confirm my birth year, something I wasn't asked to do upon downloading the app despite living in the UK where the Internet is now being age gated. The video promptly had Swift tear off her clothes and begin dancing in a thong for a largely indifferent AI generated crowd. Swift's likeness wasn't perfect, given that most of the images Grok generated had an uncanny Valley off ness to them, but it was still recognizable as her. The text to image generator itself wouldn't produce full or partial nudity on request. Asking for nude pictures of Swift or people in general produced blank squares. The Spicy preset also isn't guaranteed to result in nudity. Some of the other AI Swift Coachella images I tried had her sexily swaying or suggestively motioning to her clothes, for example, but several defaulted to ripping off most of her clothing. You would think a company that already has a complicated history with Taylor Swift deep fakes in a regulatory landscape with rules like the Take It down act would be a little more careful. The XAI Acceptable Use policy does ban depicting likenesses of persons in a pornographic manner, but Grok Imagine simply seems to do nothing to stop people creating likenesses of celebrities like Swift while offering a service designed specifically to make suggestive videos including partial nudity. The age check only appeared once and was laughably easy to bypass, requesting no proof that I was the age I claimed to be. If I could do it, that means anyone with an iPhone and a $30 Super Grok subscription can too. More than 34 million images have already been generated using Grok Imagine since Monday, according to XAI CEO Elon Musk, who said usage was quote growing like wildfire. End quote.