Welcome to Get The Gist, where every weekday we share an easy-to-read summary of the latest and greatest developments in AI—news, innovations, and trends—all delivered in under 5 minutes! ⏱

In today’s edition:

• Anthropic’s Claude AI Can Now Control Your Computer

• Google’s Gemini AI Expands Its Integrations

• Claude 3.5 Sonnet Outshines GPT-4o and Gemini

• Hugging Face Launches Open-Source ChatGPT Alternative

• And more AI news….

Top Developments

Anthropic

Anthropic’s Claude AI Can Now Control Your Computer

The Gist: Anthropic has introduced a version of its AI model, Claude, that can control computers to perform tasks like browsing the web and editing code. This marks a step towards AI agents handling everyday computer tasks, freeing up users for more creative work.

Key Details:

• Claude can now perform a range of computer tasks, such as web browsing, opening apps, and inputting text, making it more “agentic” than previous models.

• The AI's abilities include planning events, creating websites, and troubleshooting errors, although it still has a lower accuracy rate (14.9%) compared to humans.

• Companies like Canva and Replit are testing Claude to automate tasks like design and coding, but there are still concerns about errors and the model’s ability to handle complex tasks.

• Anthropic, backed by Amazon, limits risky actions, like making purchases, but the AI represents a shift towards a future where users may rely more on agents for routine tasks.

Google Gemini

Google’s Gemini AI Expands Its Integrations

The Gist: Google has expanded its Gemini AI integrations, allowing Workspace users to interact with Calendar, Keep, and Tasks. This update makes organizing and managing tasks easier by enabling actions like creating events directly from Gemini.

Key Details:

• Gemini can now create events in Calendar, make notes in Keep, and add tasks in Tasks for Google Workspace accounts managed by an admin.

• These new integrations build on previous Gemini functionality, which already allows actions in Gmail, Google Docs, and Google Drive.

• Users with Gemini Business, Enterprise, or Education add-ons can access these features if enabled by their Workspace administrator.

• Individual users with a paid Gemini Advanced plan already have similar access to AI-assisted actions across personal Google apps.

Anthropic

Anthropic's Claude 3.5 Sonnet Outshines GPT-4o and Gemini in AI Benchmarks

The Gist: Anthropic's newly updated Claude 3.5 Sonnet model delivers major performance boosts, especially in coding tasks, surpassing both OpenAI’s GPT-4o and Google’s Gemini 1.5 Pro. Additionally, the Claude 3.5 Haiku model offers impressive AI advancements at an affordable price.

Key Details:

• Claude 3.5 Sonnet has improved scores across various benchmarks, including SWE-bench (49%) and TAU-bench (69.2%), setting industry records.

• The Claude 3.5 Haiku model outperforms previous models like Claude 3 Opus and GPT-4 Turbo, with 40.6% on SWE-bench Verified.

• US and UK AI Safety Institutes verified the safety of Claude 3.5 Sonnet under their ASL-2 Standard.

• Both models are now available via Anthropic API, Amazon Bedrock, and Google Cloud, with the Haiku model's image support coming soon.

Quick Gist

NVIDIA has rebranded its upcoming AI computing GPUs to the B300 "Blackwell Ultra" series, featuring enhanced HBM3e specs and set to launch between Q2 and Q4 of 2025 (Read More).

Midjourney is introducing new web-based editing features that allow users to manipulate uploaded images with generative AI tools, marking a shift from its previous Discord-only setup (Read More).

Stability AI released Stable Diffusion 3.5, an updated open-source image generator that improves prompt adherence and diversity in outputs, addressing previous flaws (Read More).

Qualcomm and Google have partnered to enable car manufacturers to create custom AI voice assistants using a new development system based on Android Automotive OS and Qualcomm's Snapdragon Digital Chassis (Read More).

OpenAI appointed Dr. Aaron Chatterji as its first Chief Economist to study AI's economic impacts and promote its benefits across the economy, while partnering with the Lenfest Institute to fund AI initiatives in local newsrooms (Read More).

OpenAI and Microsoft are funding newsroom projects with grants up to $10 million to promote AI integration, collaborating with several publications to develop sustainable AI initiatives, despite ongoing legal challenges (Read More).

OpenAI also appointed Scott Schools as Chief Compliance Officer and Dr. Aaron Chatterji as Chief Economist, to strengthen its regulatory and economic strategy amid recent executive departures (Read More).

HuggingChat, an open-source alternative to ChatGPT developed by Hugging Face, offers users the flexibility to choose from various AI models without restrictions or payment barriers (Read More).

Meta is testing facial recognition technology to combat scam ads that use public figures' likenesses, allowing users to verify their identities and control their accounts (Read More).

THAT’S IT FOR TODAY!