Amazon expects generative AI to jolt its middle-aged Alexa voice assistant to new life.

Why it matters: The giant online seller sees the AI boom as a chance to dust off the more than half a billion devices that have Alexa built in and give them fresh relevance as delivery channels for the equivalent of a voice-enabled ChatGPT — and more.

Driving the news: Amazon has pitched its web services as an ideal place to host and run generative AI applications, but has been relatively quiet about offering services directly to consumers. That's about to change, Axios has learned.

• The first visible signs of this effort are expected to turn up when Amazon announces new devices at a Sept. 20 event at its HQ2 offices in Arlington, Virginia. However, more complete integrations will take until next year and beyond.
• "Generative AI is something we’ve been working on for a while, and it has huge potential — especially in the home," Amazon senior vice president of devices and services Dave Limp told Axios. "The opportunity to bring generative AI to customers at that scale is incredibly exciting, but we also know we need to hold a high bar as we build new features and experiences."

Between the lines: Amazon sees Alexa's presence in the home — and its existing ties to a range of other devices — as a key advantage over rivals, who are mainly focused on apps for computers and phones.

• Many of Amazon's devices have additional sensors, such as cameras and microphones, that can add context to generative AI queries.
• Alexa also supports a range of other hardware, including the Eero home network and Ring security systems Amazon sells, as well as a broad range of smart devices made by others.

These devices open the door for generative AI queries that not only answer questions, but take action.

• One could imagine, say, being able to tell a voice assistant, "I'm going on vacation for a week. Please set my devices to make sure the house is safe and we aren't using more electricity than necessary."
• Amazon's goal is to transform Alexa into an assistant that is more conversational and less transactional in nature.

Yes, but: This potential advantage is also a challenge when it comes to security and privacy. Amazon needs to be sure that generative AI systems — known to "hallucinate" by making things up — don't start opening doors, starting cars or ordering products when they shouldn't.

There are other technical hurdles. Answering queries with generative AI, for example, takes more time and computing resources than simpler responses. Amazon is keen to make sure the lag isn't too noticeable.

• Summoning Alexa in the early days of the Echo product line was a sluggish affair. Amazon has worked to reduce this "latency" and isn't keen to repeat it.

Flashback: If this sounds a bit like history repeating itself, it is. Amazon shifted its hardware focus to the home after largely missing out on the smartphone revolution with its lone hardware attempt — the Fire Phone — turning out to be an epic failure.

The big picture: Amazon's AWS has been active on the generative AI front, looking to host a wide range of uses. Amazon CEO Andy Jassy has indicated a broader interest in adopting the technology more widely, but has been short on specifics.

• Microsoft, Google and Meta have all been aggressively deploying generative AI, while Apple, which Bloomberg reports is testing a GPT-like tool internally, has remained relatively quiet.

Four leading AI providers are launching the Frontier Model Forum — an industry-led body aiming to ensure "safe and responsible development" of the most powerful AI models, Axios' Ryan Heath reports.

• Anthropic, Google, Microsoft and OpenAI are the forum's founding members, but are inviting competitors and civil society organizations to partner.

Why it matters: The companies are among seven which last Friday committed in a deal with the White House to minimize AI risks, conduct more research on AI safety and share safety best practices, among other promises.

• Critics pointed to the absence of concrete targets or measurable outcomes in the companies' commitments.

The details: The forum, per a Google spokesperson, will start with a research focus and work to develop standardized best practices for evaluating frontier model risks and to issue a "public library of solutions."

• Frontier models are "those that exceed the capabilities currently present in the most advanced existing models, and can perform a wide variety of tasks," per a statement from the companies.
• The forum will, over time, develop a formal charter, appoint an advisory board and fund collaboration with civil society groups, its founders say.

There's a global angle, with forum members promising to plug into G-7, OECD and U.S.-EU Trade and Technology Council processes, while also supporting work by the Partnership on AI and MLCommons.

Continued growth in its cloud business and strong demand for generative AI features helped Microsoft's quarterly sales and earnings exceed Wall Street expectations, the software giant said Tuesday.

What they're saying: "Organizations are asking not only how, but how fast, they can apply this next generation of AI to address the biggest opportunities and challenges they face — safely and responsibly," Microsoft CEO Satya Nadella said in a statement.

By the numbers: Revenue for the three months ended June 30 was $56.2 billion, up 8% from the prior year and more than the $55.5 billion average analysts' estimate.

• Net income was $20.1 billion, up 20% from a year ago.

Catch up quick: Microsoft has infused generative AI throughout its product line, including its Bing search engine and Edge browser as well as by adding AI-powered "copilots" to Windows, Office and other major software franchises.

• Microsoft said last week that it will charge $30 per person per month to businesses that want to use the copilots for Office apps such as Word, Excel and PowerPoint.

Separately: As part of its earnings report, Google parent Alphabet announced it plans to build generative AI features into Android 14, the next version of its mobile operating system.

• Alphabet also promoted longtime chief financial officer Ruth Porat to president and chief investment officer. She will oversee the company's "other bets" and Alphabet will search for a new CFO.

• Samsung unveiled its latest crop of foldable phones, the Galaxy Flip 5 and Galaxy Fold 5, at an event in South Korea. It also announced updated watches and tablets. (Wired)
• OpenAI released its Android app for ChatGPT on Tuesday. It has also pulled the plug on an online tool that aimed (but often failed) to distinguish which text was created by humans and which by an AI system. (Decrypt)
• President Biden Tuesday nominated Harry Coker, an ex-NSA and CIA official, as national cyber director, after ruling out acting director Kemba Walden. (Axios)
• The Senate moved Tuesday to require U.S. investors to notify the Treasury Department when they take stakes in Chinese companies in certain sectors, such as artificial intelligence and quantum computing. (Axios)
• On tap: Amazon's AWS unit is expect to introduce several new capabilities at an event later today in New York.
• A bipartisan pair of senators is urging the DOJ to devote more resources to combat AI-generated child sexual abuse material, Axios Pro's Maria Curi scooped earlier today.
• A new startup called Channel 1 plans to create AI-generated newscasts, with digital anchors, that can be customized to fit viewers' political slant. (Axios Pro)

I'm not sure if AI came up with this idea or not, but French's mustard-flavored Skittles are a thing — at least as a short-term promotion for National Mustard Day on Aug. 5.