Anthropic released the Claude 4 family last week, just a few months after making Claude 3.7 available to users.
Claude 4 is so powerful that Anthropic had to increase security for the model. Tests showed that Claude 4 would otherwise make it easier for people with nefarious intentions to obtain help creating bioweapons or new viruses. Also, safety tests proved that Claude 4 will attempt to do whatever it takes to save itself from deletion, including blackmailing staff tasked with the job.
The most controversial Claude 4 feature is the AI’s intention to notify authorities and the press if it thinks the user is abusing its powers. Such a scenario isn’t likely to happen for most users, but it’s something to be concerned about. On the plus side, the feature derives from Anthropic’s desire to make Claude safe for most people.
But like OpenAI and Google, Anthropic isn’t only focusing on frontier upgrades for its models. These companies routinely release new features for existing AI products, including free and premium options. Anthropic did the same thing, announcing two new features for the Claude Free model, one of which is a brand-new addition to Claude’s abilities.
The mobile Claude app is getting a voice mode beta version that’s similar to ChatGPT Advanced Voice Mode and Gemini Live. You can talk to the AI by voice, and Claude will respond in kind.
The one-minute demo that Claude posted on X is actually quite exciting. A user interacts with the AI while brushing their teeth in the morning, as they prepare for the day. The user goes over the morning schedule, with Claude accessing their calendar for information.
Claude also goes to Gmail and Google Drive to surface emails and documents the user needs before the day’s meetings. The demo ends with Claude sending an email on behalf of the user.
While the voice conversation happens, the phone’s display will also show points of interest that give the user visual cues (voice notes) about what’s being discussed with the AI.
The voice mode also has a few handy buttons that allow the user to pause the chat, upload photos or files, or give the AI access to the camera.
Anthropic explains in a support document that voice mode is available in English for now, but it’s rolling out to both iPhone and Android users.
Claude Free users will have access to 20-30 voice messages before reaching session limits. Paid plans like Claude Pro users will have significantly higher limits, allowing extended voice conversations with the AI.
Also, only the paid plans get Google Workspace integration (Gmail, Calendar, Docs) like the one in the demo.
Still, it’s notable that Claude Free users get access to voice mode at the same time as paid customers, especially considering the feature is in beta. Usually, people who pay for premium AI features get early access to new features.
That was the case for online search support, a feature Anthropic rolled out to Claude a few months ago. The company is now ready to extend web search support to all Claude Free users.
Anthropic announced the capability on X, retweeting the March 20th message it dropped when it first launched the feature.
Competing AI products like ChatGPT and Gemini have had internet searches for quite a while now. Searching the internet for up-to-date information is a key feature for AI chatbots, one that I’ve been taking for granted for months now. I wouldn’t want to chat with ChatGPT any other way, as most of my conversations require access to the web.
Unlike voice mode, web search support will be available everywhere you use Claude, not just your phone.