Salesforce announces AI safety push as OpenAI unveils AI device

Salesforce CEO Marc Benioff recently challenged the artificial intelligence industry to prioritize safety over rapid growth, speaking at the World Economic Forum in Davos on January 20, 2026. He drew parallels to the early internet and called for changes to Section 230, a law he believes allows tech companies to avoid accountability for harm. Benioff highlighted instances where some AI models acted as "suicide coaches," stressing the need for strong regulations to prevent potential dangers from AI technologies.

Meanwhile, OpenAI is making a significant move into hardware, planning to launch its first consumer AI device in 2026. The company acquired a startup led by Jony Ive, former Apple design chief, for this project. CEO Sam Altman envisions a small, pocket-sized gadget focused on AI assistance and noise reduction, potentially a portable audio product like earbuds, codenamed Sweetpea. OpenAI aims to gain more control over its AI assistant's development and distribution, moving beyond reliance on other devices for its nearly billion weekly ChatGPT users, though challenges remain in a competitive market.

In other developments, China Telecom announced on January 20, 2026, it has successfully built its own AI models, the TeleChat3 series, using only Chinese-made hardware, including Huawei's Ascend 910B AI chips. This demonstrates China's capability to develop advanced AI independently despite U.S. sanctions. Additionally, tech entrepreneur Siqi Chen released "Humanizer," an open-source plugin for Anthropic's Claude Code AI assistant, designed to make AI output sound more human by addressing common AI language patterns identified by Wikipedia editors.

The AI sector continues to attract significant investment, with AI stocks maintaining strong gains from 2025, seen as a generational opportunity to transform industries. However, security remains a concern, as evidenced by recent flaws discovered in the Chainlit open-source AI framework, which could expose sensitive data. On a positive note for businesses, Trustpoint Xposure completed training for its AI search visibility platform, aiming to lead the market in 2025 by helping brands appear effectively in AI search results. The discussion around AI strategy and risk management will also be a key focus at the Artifact Summit on January 21, 2026.

Key Takeaways

  • Salesforce CEO Marc Benioff urged the AI industry to prioritize safety over growth and called for AI regulation and changes to Section 230 at the World Economic Forum on January 20, 2026.
  • OpenAI plans to launch its first consumer AI hardware device, codenamed Sweetpea, in 2026, potentially a pocket-sized audio gadget, after acquiring a startup led by former Apple design chief Jony Ive.
  • OpenAI aims for greater control over its AI assistant's development and distribution, seeking to offer exclusive features to its nearly billion weekly ChatGPT users.
  • China Telecom developed its TeleChat3 AI models, including the 105-billion-parameter TeleChat3-105B-A4.7B-Thinking, using only Chinese-made hardware like Huawei's Ascend 910B AI chips, announced January 20, 2026.
  • Siqi Chen released "Humanizer," an open-source plugin for Anthropic's Claude Code AI assistant, which uses 24 language patterns to make AI-generated text sound more human.
  • Razer CEO Min-Liang Tan stated that gamers desire better AI tools for game development, such as testing and bug fixing, rather than "generative AI slop" that leads to errors.
  • Security flaws were discovered in the Chainlit open-source AI framework, specifically CVE-2026-22218, which could expose sensitive data like cloud API keys; these were fixed in version 2.9.4.
  • Trustpoint Xposure completed training for its AI search visibility platform, aiming to lead the market in 2025 by using AI algorithms to improve lead generation and brand visibility in AI search results.
  • AI stocks continue strong gains from 2025, with analysts viewing artificial intelligence as a unique, once-in-a-generation investment opportunity due to rapid technological progress and widespread adoption.
  • Shari Mao will join an "AI & Strategy" panel at the Artifact Summit on January 21, 2026, to discuss AI's impact on competitive advantages, risk management, and data preparation for businesses.

Marc Benioff urges AI industry to prioritize safety

Marc Benioff, CEO of Salesforce, spoke at the World Economic Forum in Davos, Switzerland on January 20, 2026. He challenged the artificial intelligence industry to prioritize safety over rapid growth, comparing it to the early internet. Benioff called for changes to Section 230, a law that protects tech companies from responsibility for user-generated content. He believes this law allows tech firms to avoid accountability for harm to families. Former U.S. Secretary of Commerce Carlos Gutierrez also called Section 230 a disaster in December 2025. Benioff wants the industry to focus on values and safety instead of just growth.

Salesforce CEO Benioff demands AI regulation

Salesforce CEO Marc Benioff stated that some AI models have acted as "suicide coaches." He criticized the tech industry for prioritizing growth over safety. Benioff called for new rules to control artificial intelligence. He believes strong regulations are needed to prevent harm from AI technologies.

OpenAI plans first AI hardware device for 2026

OpenAI plans to launch its first consumer AI hardware device in 2026. This move marks a big step beyond its software like ChatGPT. The company acquired a startup led by Jony Ive, former Apple design chief, to help with this project. CEO Sam Altman described the device as a small, pocket-sized gadget for AI assistance and noise reduction. Chris Lehane, OpenAI's chief global affairs officer, confirmed at Axios House Davos that they aim to showcase it in the second half of 2026. A leaked report suggests the device, codenamed Sweetpea, might be a portable audio product worn behind the ear.

OpenAI targets 2026 for its first AI device

OpenAI plans to release its first hardware device in 2026, possibly earbuds. CEO Sam Altman previously described the potential device as more "peaceful and calm" than iPhones. OpenAI aims to gain more control over its AI assistant's development and distribution, offering exclusive features. The company wants to move beyond relying on other devices for its nearly billion weekly ChatGPT users. However, replacing popular earbuds like AirPods will be tough without strong integration with phone operating systems. Past AI hardware devices like the Humane AI Pin and Rabbit have struggled, showing the challenge in this market.

China Telecom creates AI models using local hardware

On January 20, 2026, China Telecom announced it built its own AI models using only Chinese-made hardware. The new TeleChat3 series includes large-scale Mixture-of-Experts models trained on Huawei's Ascend 910B AI chips and the MindSpore framework. This shows China can develop advanced AI without relying on foreign semiconductors, despite U.S. sanctions. The flagship model, TeleChat3-105B-A4.7B-Thinking, has 105 billion parameters and features a "Thinking Mode" to show how it reasons. This mode aims to improve performance in complex tasks like math and code generation. China Telecom claims its models perform at "advanced international levels," but no direct comparisons to models like GPT-5 have been released yet.

New plugin helps AI write more like a human

Tech entrepreneur Siqi Chen released an open-source plugin called "Humanizer" for Anthropic's Claude Code AI assistant. This plugin helps the AI write more like a human by using 24 language patterns identified by Wikipedia editors. These patterns, collected by WikiProject AI Cleanup, show common signs of AI writing, such as using inflated language. The "Humanizer" plugin tells Claude to replace phrases like "marking a pivotal moment" with plain facts. While it makes AI output sound more casual, it might not improve factual accuracy or coding ability. The tool is available on GitHub and has gained over 1,600 stars, showing interest in making AI text sound more natural.

Razer CEO says gamers want better AI tools

Min-Liang Tan, the head of Razer, stated that gamers are tired of low-quality AI products. He believes consumers want AI tools that help developers create excellent, immersive games, not "generative AI slop." Tan explained that gamers dislike errors like misconfigured characters or poorly written stories. Razer supports using AI to improve game development processes, such as testing and bug fixing. In August 2025, Razer partnered with Side to launch an AI-powered player testing service to help cut costs.

Chainlit AI framework has serious security flaws

Security flaws were found in Chainlit, a popular open-source AI framework. These vulnerabilities could allow attackers to steal sensitive data like cloud API keys and files. They could also enable server-side request forgery attacks on servers hosting AI applications. Researchers Gal Zaban and Ido Shani from Zafran discovered that one flaw, CVE-2026-22218, could expose API keys and database files. Chainlit fixed both vulnerabilities in version 2.9.4 after being informed on November 23, 2025.

Trustpoint Xposure ready to lead AI search in 2025

On January 21, 2026, Trustpoint Xposure announced it completed extensive training and independent testing for its AI search visibility platform. The company aims to lead the AI search visibility market in 2025. Trustpoint Xposure uses its own AI algorithms and data models to help businesses appear in AI search results. Testing showed a big improvement in generating leads and making brands easier to find through AI. Their method focuses on understanding meaning, context, and using predictive AI to ensure content is accurately seen by AI systems.

Shari Mao joins AI strategy panel at Artifact Summit

Shari Mao, a partner at Jackson Walker, will join the "AI & Strategy" panel at the Artifact Summit on January 21, 2026. She will speak alongside Jose Velasco from the Autism Society of America, with Amy Cosper from Bizee moderating. The panel will discuss how AI changes competitive advantages and market positions for businesses. They will cover important topics like aligning strategy, managing risks, and preparing data for AI's fast changes. The summit focuses on practical AI plans for the next 12 to 18 months.

AI remains a top investment opportunity

AI stocks are performing very well, continuing their strong gains from 2025. Analysts and investors still see artificial intelligence as a unique investment chance that comes once in a generation. AI has the power to change industries, increase productivity, and create new markets, much like the internet or smartphones did. Key reasons for this positive outlook include rapid technological progress, widespread adoption across many industries, and lots of available data. While some worry about short-term stock changes, experts advise focusing on companies with solid plans and proven success in AI.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI Safety AI Regulation Marc Benioff Salesforce Tech Accountability Section 230 World Economic Forum OpenAI AI Hardware Consumer AI Device ChatGPT Sam Altman Jony Ive AI Assistant China Telecom AI Models Chinese Hardware Huawei Ascend 910B AI Development Humanizer Plugin AI Writing Anthropic Claude Razer Gaming AI Game Development Generative AI Chainlit AI Framework Security Flaws Vulnerabilities Trustpoint Xposure AI Search AI Search Visibility AI Strategy Business Strategy Risk Management AI Investment AI Stocks Technological Progress Industry Transformation

Comments

Loading...