Latest news with #languageModels
WIRED
2 days ago
- Business
- WIRED
OpenAI Just Released Its First Open-Weight Models Since GPT-2
Aug 5, 2025 1:00 PM The models, gpt-oss-120b and gpt-oss-20b, represent a major shift for the AI company. Photograph:OpenAI just dropped its first open-weight models in over five years. The two language models, gpt-oss-120b and gpt-oss-20b, can run locally on consumer devices and be fine-tuned for specific purposes. For OpenAI, they represent a shift away from its recent strategy of focusing on proprietary releases, as the company moves towards a wider, and more open, group of AI models that are available for users. "We're excited to make this model, the result of billions of dollars of research, available to the world to get AI into the hands of the most people possible," said OpenAI CEO Sam Altman in an emailed statement. Both gpt-oss-120b and gpt-oss-20b 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 GPT-2, 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, cofounder 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,' 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 GB of memory. The two new models from OpenAI 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 model releases from Alibaba's Qwen 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. In addition to the evaluations OpenAI typically runs on its proprietary models, the startup customized the open-weight option to see how it could potentially be misused by a 'bad actor' who downloads the tool. 'We actually fine-tuned the model internally on some of these risk areas,' said Eric Wallace, a safety researcher at OpenAI, 'and measured how high we could push them.' In OpenAI's tests, the open-weight model did not reach a high level of risk, as measured by its preparedness framework. 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 o4-mini models, which are proprietary, and even out-performing 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. At the beginning of this year, the Chinese startup DeepSeek stunned Silicon Valley with the release of its cheap-to-run model that was open-weight. While the release blog about gpt-oss does not mention DeepSeek or any other Chinese AI company directly, Altman is clear that he wants innovation around open-weight models to happen in the United States. "Going back to when we started in 2015, OpenAI's mission is to ensure AGI that benefits all of humanity,' said Altman in a statement. 'To that end, we are excited for the world to be building on an open AI stack created in the United States, based on democratic values, available for free to all and for wide benefit." In the US, the open-weight leader has been Meta. The tech giant released the first of its Llama series of models back in 2023, with Meta's most recent release, Llama 4, arriving a few months ago. With that in mind, Meta is currently hyper-focused on building AI that can surpass human cognition, often called superintelligence by AI insiders. The company recently launched a new, internal lab focused on this lead by Alexandr Wang, the former CEO of Scale. Mark Zuckerberg has signaled that the company may move away from open-source for future models, citing potential safety concerns. The gpt-oss release also comes as the AI talent war between companies, like OpenAI and Meta, continues to ramp up. In 2025, AI researchers who have in-demand talents are being presented with astronomical offers to switch companies. The latest releases from OpenAI could be stiff competition for Meta, depending on how the gpt-oss models are received by developers.
Yahoo
05-07-2025
- Yahoo
X brings AI into Community Notes to fight misinformation at scale humans can't match
X, formerly known as Twitter, has started testing a new approach in its Community Notes program by bringing AI into the mix. The company has begun piloting the use of large language models (LLMs) to write contextual notes that flag or clarify potentially misleading posts. Community Notes began in 2021 as a user-driven initiative to combat misinformation. Volunteers could add notes under posts that might be deceptive or misinterpreted, such as labeling AI-generated videos to prevent people from believing false events. Human raters would then decide which notes were helpful. Only the most helpful notes would appear publicly. The latest update adds AI-generated notes to this process, but with a key safeguard: only humans can rate which notes are ultimately shown. 'Allowing automated note creation would enable the system to operate at a scale and speed that is impossible for human writers,' the researchers write. They believe LLMs could 'potentially provide context for orders of magnitude more content across the web.' The AI model will continue to evolve through a method called reinforcement learning from community feedback (RLCF). The process refines note generation by using responses from people with a wide range of perspectives. The result, researchers say, should be notes that are more 'accurate, unbiased and helpful.' Despite the promise of scale, researchers highlight several risks. AI-generated notes can sound persuasive even when they're wrong, a known issue with many LLMs. There's also a risk of producing too-similar notes, which could dilute the diversity of perspectives the system aims to include. Another concern is that human contributors may participate less if AI generates too many notes. Human raters could also become overwhelmed trying to assess the volume of AI submissions. The study lays out several potential solutions. One idea is to develop AI co-pilots that assist human writers with research, speeding up their output. Another is to use AI tools to help raters evaluate notes more efficiently. To preserve quality, researchers also suggest vetting human contributors more thoroughly and customizing LLMs for note-writing. They propose adapting validated notes for reuse in similar cases, saving time and reducing redundancy. Despite increasing automation, researchers stress the continued importance of human oversight. 'The goal is not to create an AI assistant that tells users what to think,' the study states, 'but to build an ecosystem that empowers humans to think more critically and understand the world better.' X's move comes as misinformation continues to spread rapidly across social media platforms, often outpacing efforts to correct it. By combining human insight with AI's processing power, the company hopes to better manage false narratives in real time. The pilot will help determine whether this hybrid approach can maintain trust while boosting coverage. The study is available on arXiv.
Wall Street Journal
01-07-2025
- Business
- Wall Street Journal
Hollywood Confronts AI Copyright Chaos in Washington, Courts
Natasha Lyonne stayed up all night furiously texting and calling Cate Blanchett, Ron Howard and everyone else she knows in Hollywood, asking them to sign her letter to the Trump administration. The White House is about to issue its artificial-intelligence action plan, a document that could influence how U.S. copyright rules are applied to training large language models. Tech companies say they need the material to train their models and keep up with China in an AI race with grave national-security implications.
Yahoo
30-06-2025
- Business
- Yahoo
Apple May Use Outside AI to Power Siri in Major Reversal
Apple Inc. is considering using artificial intelligence technology from Anthropic PBC or OpenAI to power a new version of Siri, sidelining its own in-house models in a potentially blockbuster move aimed at turning around its flailing AI effort. The iPhone maker has talked with both companies about using their large language models for Siri, according to people familiar with the discussions. But no agreements have been reached. Bloomberg's Mark Gurman reports. 擷取數據時發生錯誤 登入存取你的投資組合 擷取數據時發生錯誤 擷取數據時發生錯誤 擷取數據時發生錯誤 擷取數據時發生錯誤
Bloomberg
30-06-2025
- Business
- Bloomberg
Apple Considers Using Outside AI to Power Siri in Major Reversal
Apple Inc. is considering using artificial intelligence technology from Anthropic PBC or OpenAI to power a new version of Siri, sidelining its own in-house models in a potentially blockbuster move aimed at turning around its flailing AI effort. The iPhone maker has talked with both companies about using their large language models for Siri, according to people familiar with the discussions. It has asked them to train versions of their models that could run on Apple's cloud infrastructure for testing, said the people, who asked not to be identified discussing private deliberations. If Apple ultimately moves forward, it would represent a monumental reversal. The company currently powers most of its AI features with homegrown technology that it calls Apple Foundation Models and had been planning a new version of its voice assistant that runs on that technology for 2026. (Source: Bloomberg)
