Meta’s Threads is enhancing its recently introduced Live Chats feature and broadening user access, the company revealed on Tuesday. The enhancements encompass translation support, new tools for chat organizers, and additional functionalities.

With translation support, chat discussions become more accessible for users globally. Threads is also extending the option to initiate Live Chats to all “Community Champions,” defined by the company as users who enjoy significant followings within their communities, frequently engage in those communities, and maintain active conversations.

Hosts can now invite up to three co-hosts into their Live Chat to facilitate easier management of discussions. Threads states this new feature is akin to having a guest on your show or an additional voice to guide the conversation. Additionally, hosts are now able to delete messages for all participants, and the platform is exploring methods to enhance the visibility of host messages in chats.

Upon its initial launch, Threads faced challenges in competing with X for real-time discussions, lacking essential features such as strong search functionality, hashtags, and a chronological feed. Since then, Threads has integrated these features and is now further setting itself apart with Live Chats, designed for immediate engagement that even X lacks.

The concept behind Live Chats is to enhance Threads’ timeliness and relevance. Since the feature’s introduction, Threads reports hundreds of chats hosted nearly every day, with thousands of users participating. The features announced today respond to creator feedback, Threads indicates.

Through Live Chats, users can share messages, photos, videos, links, and emoji reactions. Up to 150 participants can engage actively within a chat. Once this cap is reached, additional users can still view the chat, react to messages, and take part in polls in “spectator” mode.

Threads also hinted today that desktop support is on the way and that pinned messages are being developed, both of which have been highly requested by users.

Earlier this month, Threads achieved 500 million monthly active users, nearly three years after it was launched as a rival to X. Over the past year, Threads has rolled out various new features such as DMs, ghost posts, and desktop messaging, contributing to the platform’s growth.

Jeff Bezos’ space venture, Blue Origin, is recommitting to utilizing its New Glenn mega-rocket once more this year following last month’s detonation. However, the company still remains uncertain about the initial cause of that explosion.

In his most comprehensive public disclosure regarding the explosion to date, CEO Dave Limp indicated on Blue Origin’s website on Tuesday that the company is still endeavoring to “pinpoint and rectify the root cause” of the incident. “Initial findings suggest the aft region of the first stage” of the rocket, Limp noted, adding that the firm is relying on “detailed data from numerous camera views and sensors.”

After dedicating over a decade to the careful and methodical development of New Glenn, Blue Origin had accelerated its pace with the rocket. Its maiden flight took place in January 2025, and New Glenn was gearing up for its fourth attempt when it detonated during tests on May 28 this year. (Fortunately, no one was injured in the explosion.)

Blue Origin is eager to get the rocket back into operation as promptly as possible due to the company’s pivotal role in NASA’s initiative to send humans back to the moon before President Trump completes his term.

To achieve this, Blue Origin must not only identify and rectify what triggered last month’s explosion but also reconstruct its launchpad in Cape Canaveral, Florida, as this is currently the company’s sole launchpad capable of accommodating the colossal rocket.

In his statement, Limp commended Blue Origin’s staff for rapidly advancing efforts in this area.

The company sustained the loss of a lightning tower in the explosion, along with the significant piece of machinery that transports New Glenn to the launchpad and erects it (known as the transporter-erector). Nearby structures on site also suffered damage from the blast. Nonetheless, Limp remarked that “we experienced many fortunate circumstances, too, and aim to capitalize on them.”

Some of those fortunate circumstances include the condition of the on-site water tower, gas tanks, and rocket integration facility, all of which are “in good condition.”

Limp provided additional insights on how Blue Origin plans to modify its existing launch complex in Cape Canaveral. The company will forgo the transporter-erector method and will utilize a substantial crane to elevate New Glenn on the launchpad prior to launches.

This strategy would not only enable Blue Origin to resume flights sooner than anticipated, but it would also boost New Glenn’s flight frequency, according to Limp. Prior to the May explosion, Blue Origin had envisioned conducting as many as 12 launches this year.

Tesla has initiated testing of a production model of its Cybercab featuring two seats, devoid of a steering wheel and pedals, in Austin, Texas. Currently, the tests are conducted with a safety observer seated in the right passenger position, as shown in a video shared on X, the social media site owned by Tesla’s CEO Elon Musk.

This test occurs almost two years after Tesla first unveiled the Cybercab’s design, intended to function as a fully autonomous robotaxi that can be requested via Tesla’s app. Approximately a year prior, Tesla began testing a Tesla Robotaxi service in Austin using Model Y SUVs, which, at times, included safety monitors.

In recent weeks, Tesla has been evaluating prototype variants of the Cybercab, fitted with a steering wheel and pedals, across several U.S. cities. Additionally, they have been seen parking numerous vehicles in lots in select cities, prompting speculation that the company may be on the verge of launching a genuinely expansive robotaxi network.

One of the potential obstacles to this vision is likely to be addressed soon. Last week, the National Highway Traffic Safety Administration unveiled a proposal that would not require brake pedals in “vehicles designed to be driven exclusively by automated driving systems.” The proposal is currently in the public commentary phase but is anticipated to pass later this year.

Musk and other Tesla leaders have claimed they will outpace the current robotaxi frontrunner, Waymo, for several reasons. A primary reason is that Tesla is manufacturing both the vehicles and the driving software, expected to provide significantly more control over costs compared to Waymo, which depends on partnerships with brands like Jaguar and Zeekr for its vehicles. Tesla is also attempting to ensure the Cybercab is fully autonomous solely using cameras, while Waymo employs a far more intricate array of sensors, including lidar and radar.

Tesla — along with Musk — has spent years assuring that the company would achieve a completely autonomous vehicle, yet it has yet to deploy one on a large scale. The closest it has come is the robotaxi service in Austin, which has seen fluctuations in size over the past year. Some of these vehicles have been involved in several minor accidents, with at least two caused by remote operators.

Of course, Waymo’s robotaxis have also been in minor accidents, and as that company has expanded, it has faced numerous difficulties and edge cases, many of which are still being addressed. The Alphabet-owned company’s robotaxis currently cannot utilize highways due to challenges in navigating construction zones, resulting in a recall (which aimed to avoid highways, not to solve the underlying issue). Waymo’s vehicles have also struggled to navigate around flooded areas during heavy rain, leading to another recall. Additionally, they have encountered issues with legal driving around school buses.

Tesla has faced its own challenges with the Austin robotaxi network, but some of these have gone unnoticed mainly because it employs only lightly modified versions of its consumer Model Y SUVs. Waymo’s sensor-rich I-Pace SUVs (and its light blue Zeekr vans, which are beginning to appear on roads now) are considerably easier to identify when they make mistakes.

Deploying gold-colored, two-seater Cybercabs will place an even greater emphasis on Tesla’s quest to create a nationwide robotaxi network, bringing both its challenges and achievements into sharper focus.

X is simplifying the connection for AI assistants such as Claude, Cursor, Grok Build, and other MCP-compatible applications to directly interact with the platform through a newly launched hosted MCP server.

On Monday, the social network owned by Elon Musk introduced a hosted Model Context Protocol (MCP) server that enables AI tools to communicate with the X API utilizing a user’s personal account permissions.

To provide context, MCP is an open standard that outlines a standardized method for AI models to link with external tools and services. In the past, developers aiming for an AI assistant like Claude or Cursor to connect with X had to create their own MCP server, manage hosting, link to the X API, and oversee authentication. Now, X manages the MCP, allowing users to authenticate using their own X account permissions.

This change lets developers reduce the time spent on integration tasks, enabling them to concentrate on their actual projects.

For a long time, developers have been capable of searching X, reading posts, looking up users, analyzing discussions and trends, and more using the platform’s API. The hosted MCP doesn’t introduce new functionalities; it simply makes existing ones easier to access for AI applications. By doing this, X can establish itself as an information network rich in real-time data for retrieval and analysis, rather than just a social platform.

This initiative sees X joining an increasing number of companies that now provide their own official MCP servers or endpoints, such as GitHub, Slack, Notion, Stripe, and Salesforce.

Naturally, there is always apprehension that by eliminating an infrastructure barrier, X risks increasing automated postings or spam.

It’s important to highlight that the hosted MCP does not circumvent X’s API regulations, which continue to limit its usage if the company identifies spammy conduct.

Earlier this year, X also updated its API v2 to tackle the problem of AI-generated spam, especially automated replies to conversations. Additionally, it recently revised its API pricing, raising the fee for posting messages to $0.015, and for sharing links to $0.20. The price hikes were intended to “reduce vectors of misuse,” X indicated at the time — suggesting it’s becoming increasingly costly to spam on X.

The global demand for copper is substantial, but with the shift towards renewable energy and the expansion of data centers, it’s set to increase significantly. A recent study indicates that by 2050, we must produce more copper than has ever been extracted throughout human history.

Much of this copper is utilized in the electrical grid, which is aging in the U.S.

“We’re reaching a critical point with AI and the electrification of nearly all sectors, resulting in a scenario where the energy grid is strained and overstressed,” stated Amir Mashal, founder and CEO of Arcturus, in an interview with TechCrunch.

While one approach may be to simply increase copper usage, Mashal noted that his stealth-mode startup offers a different solution. Arcturus aims to decrease the energy lost as heat in electrical conductors by integrating carbon nanomaterials into copper and aluminum through laser technology. Substituting traditional copper with Arcturus’s composite would enable equal-sized power lines to transmit more electricity.

In practical terms, this could lead to a 50% reduction in electrical grid losses, potentially freeing up around 3% more electricity on average and up to 10% more during peak congestion when the grid is under the most pressure. At the lower end, this represents approximately a year’s worth of growth in demand in the U.S.

“Copper becomes less conductive as it heats up, causing more energy to be wasted as heat the hotter it gets,” Mashal explained. “As I continued to delve deeper into the issue, everything began to resonate with me because I recognized that this limitation recurs everywhere. The contemporary world is fundamentally powered by metals.”

Although the grid is the ultimate aim for a materials startup like Arcturus, the company is initially targeting smaller applications including drones, robotics, and data centers, where even minor increases in electricity efficiency can result in significant benefits. 

The company exclusively disclosed to TechCrunch that it secured $8 million in a seed funding round led by Initialized Capital, with contributions from Toyota Ventures, Breakthrough Energy Discovery, 1517, and Wireframe Ventures.

Mashal has been quietly perfecting his materials in a garage located in Malibu, California, where he can currently manufacture several centimeters of wire as a proof of concept. With the new investment, he aims to increase production to tens of meters, enabling testing of the nano-infused materials in diverse applications, including electric motor windings and busbars for power distribution.

Although the properties of these materials are novel, Mashal stated they are designed to serve as a “drop-in replacement” for existing copper and aluminum applications. “They retain the same form factors, require no system redesign, and necessitate no new training for personnel to manage or crimp the material.” 

Materials from Arcturus could lead to lighter drones or more efficient electric vehicles. By minimizing heat loss, they can also lessen the cooling demands of data centers.

“All those sectors encounter similar bottlenecks, whether it’s a drone striving for double flight duration or a graphics card overheating,” Mashal remarked. “These are all areas where our materials can create significant disruption.”

The creators of the video and podcast recording tool Riverside are introducing a novel method for users to connect with their audiences: newsletters.

However, Riverside isn’t directly competing with established newsletter services like Mailchimp, Substack, Beehiiv, or Ghost. Instead, understanding that its userbase is already producing a significant amount of content, the company is providing its recording tool users with an AI solution to transform their existing videos and podcasts into newsletters, allowing them to send these directly from the application. Additionally, users can compose and dispatch newsletters from the ground up without utilizing the AI conversion feature.

“While platforms like Substack and Beehiiv present a blank slate, our creators and business clients are already generating rich, information-laden spoken content on Riverside. For many individuals, speaking comes more effortlessly and naturally than crafting text from scratch, and the concepts are already embedded within the conversation. Therefore, instead of urging them to begin anew in a separate tool, we assist them in converting a pre-existing recording into newsletter-ready format with considerably less effort,” stated Riverside’s co-founder and CEO Nadav Keyson to TechCrunch.

The company is further enhancing its recording suite to accommodate multi-camera recording setups and is providing users with the capability to include remote guests in their recordings.

The update also introduces new AI functionalities. Users can leverage AI to generate a preliminary draft of a recording immediately after completion, and the assistant can formulate hooks and content tailored for various social media channels. Furthermore, the company is integrating an AI video enhancement feature, designed from conversational video podcasts, which it claims can enhance the lighting, depth, and clarity of recordings.

Riverside, which has secured over $60 million in funding, joins a range of platforms that are exploring alternative publishing avenues to diversify or grow their revenue streams. For example, in March, Substack unveiled an integrated recording studio that competes head-to-head with Riverside, while Beehiiv also dipped into podcasting in April. In June, the social network Mastodon stated it would enable users to publish their posts as newsletters.

As organizations face challenges in AI integration, they are more willing to seek external assistance — leading service providers to establish new specialized groups to ensure they receive it.

On Tuesday, Amazon Web Services (AWS) introduced a new internal division dedicated to AI-targeted forward-deployed engineers. Engineers within this new unit will integrate into companies to implement custom-built agents, emphasizing rapid engagements and enabling customer autonomy.

In an announcement regarding the new organization, AWS VP of Frontier AI Francessca Vasquez highlighted that the organization would exceed the mere creation and maintenance of requested systems. “Clients exit AWS FDE deployments equipped with both innovative solutions and enhanced engineering capabilities,” the announcement states. “In addition to agentic systems operating within their own AWS environment, they acquire enduring AI competencies, workflows, and frameworks that they can leverage for independent innovation.”

Amazon has committed $1 billion to this new organization, although this amount pertains to internal Amazon assets rather than a joint venture or standard investment. 

The forward-deployed engineer (FDE) model, first developed by Palantir, has gained traction as a method for managing AI implementations. In a standard FDE setup, an engineer from the contracting firm (in this instance, AWS) temporarily assists the client during the system establishment, allowing for immediate responses to internal opportunities or issues as they arise. 

Within the FDE framework, much of the applicable technology can be reused across different deployments while still being customized to fit each company’s distinct requirements and workflows. It additionally provides the client organization with a surge of expertise and places the primary responsibility for deployment with the contractor. The main drawback is the labor requirement, as it necessitates maintaining a full team of FDE engineers to install and manage the organization’s technology.

Both OpenAI and Anthropic have recently initiated their own FDE partnerships, valued at $4 billion and $1.5 billion, respectively. In these instances, the AI laboratories collaborated with private equity firms, which supplied the funding needed to launch and facilitated connections with corporate clients within their portfolios.

The timer has begun. In just a matter of months, more than 10,000 innovators, entrepreneurs, financiers, and industry authorities will converge in San Francisco for TechCrunch Disrupt 2026. Here’s your exclusive chance: From October 10-16, host a Side Event and take charge of the space.

Why this moment? Why you?

By hosting a Side Event, you seize control — placing your brand in the spotlight with those who are most impactful. Whether you’re a VC eager to initiate discussions, a firm poised to present groundbreaking developments, or a community leader with a message to share, this is your opportunity to:

• Connect directly with thousands of the brightest minds in technology.
• Forge genuine relationships with potential collaborators, investors, and partners.
• Control your narrative and highlight your expertise on your terms.
• Create excitement that resonates well beyond the actual event.

What can you organize?

Anything that aligns with your ambitions. Industry discussions. VC consultation hours. In-depth workshops. Relaxed networking socials. Exclusive dinners. The structure is up to you — we ask only that you bring the enthusiasm.

Your chance to create something special

Side Events will occur from October 10-16, with one key advantage: Evening events scheduled during October 13-15 (the conference period) are highly recommended — fewer competing events lead to more significant conversations and a focused audience.

The way ahead

Submit your event proposal by September 4 — share your vision, objectives, and logistical needs. Our team will take care of the rest. Complete support. Total visibility. No costs to apply or engage.

The chance won’t linger

Apply today and secure your place at the leading tech event of 2026.

Last year, Proton, the company dedicated to enhancing user privacy in productivity apps, introduced a public AI chatbot named Lumo. An upgrade for the chatbot was released on Tuesday.

Lumo 2.0 introduces a range of new features, including the ability to recognize and generate images. Users can now upload images into Lumo and utilize the chatbot to analyze or modify them. Like other large language models, Lumo can also create visuals based on user prompts.

The 2.0 update also enhances Lumo’s functions for Projects — the feature that enables users to upload files and work via Proton’s additional services such as email and cloud storage. Projects now come equipped with user-controlled persistent memory, allowing Lumo to remember a user’s preferences across different conversational interactions.

Furthermore, the update significantly boosts Lumo’s performance compared to its earlier version. According to the company, the 2.0 version processes most inquiries up to 76% faster than its predecessor. It also introduces a new “thinking mode” designed for handling more intricate questions or problems.

“Lumo 2.0 has been completely revamped, and the addition of thinking mode empowers it with robust new features,” stated Andy Yen, founder and CEO of Proton. “Lumo 2.0 exemplifies that users can have both advanced AI functionalities and substantial privacy safeguards.”

In terms of utility, the public version of Lumo is comparable to other leading chatbots. It responds to inquiries in a manner similar to Gemini and ChatGPT, offering roughly the same detail and context.

However, Proton sets Lumo apart from other chatbot services by emphasizing its privacy measures. It utilizes what it describes as zero-access encryption architecture, which ensures that users’ data is encrypted during transmission and storage, accessible solely to the user. The company also asserts that no server-side session logs are maintained, meaning no one at Proton can access the content of conversations. Proton further commits to never using customer data for AI training or sharing it with third-parties.

Lumo 2.0 is now available to users. Alongside the free public version, Proton has paid tiers (Plus and Professional) that offer users significantly greater access and features.