• The "Godfather of AI" outlined why he thinks Google moved slower on rolling out its chatbots.
• Geoffrey Hinton, who spent more than a decade at Google, said reputational considerations were key.
• Hinton, who focuses on AI safety, said he couldn't comment on if Sam Altman has a "good moral compass."

When it comes to winning the AI race, the "Godfather of AI" thinks there's an advantage in having nothing to lose.

On an episode of the "Diary of a CEO" podcast that aired June 16, Geoffrey Hinton laid out what he sees as a key difference between how OpenAI and Google, his former employer, dealt with AI safety.

"When they had these big chatbots, they didn't release them, possibly because they were worried about their reputation," Hinton said of Google. "They had a very good reputation, and they didn't want to damage it."

Google released Bard, its AI chatbot, in March of 2023, before later incorporating it into its larger suite of large language models called Gemini. The company was playing catch-up, though, since OpenAI released ChatGPT at the end of 2022.

Hinton, who earned his nickname for his pioneering work on neural networks, laid out a key reason that OpenAI could move faster on the podcast episode: "OpenAI didn't have a reputation, and so they could afford to take the gamble."

Talking at an all-hands meeting shortly after ChatGPT came out, Google's then-head of AI said the company didn't plan to immediately release a chatbot because of "reputational risk," adding that it needed to make choices "more conservatively than a small startup," CNBC reported at the time.

The company's AI boss, Google DeepMind CEO Demis Hassabis, said in February of this year that AI poses potential long-term risks, and that agentic systems could get "out of control." He advocated having a governing body that regulates AI projects.

Gemini has made some high-profile mistakes since its launch, and showed bias in its written responses and image-generating feature. Google CEO Sundar Pichai addressed the controversy in a memo to staff last year, saying the company "got it wrong" and pledging to make changes.

The "Godfather" saw Google's early chatbot decision-making from the inside — he spent more than a decade at the company before quitting to talk more freely about what he describes as the dangers of AI. On Monday's podcast episode, though, Hinton said he didn't face internal pressure to stay silent.

"Google encouraged me to stay and work on AI safety, and said I could do whatever I liked on AI safety," he said. "You kind of censor yourself. If you work for a big company, you don't feel right saying things that will damage the big company."

Overall, Hinton said he thinks Google "actually behaved very responsibly."

Hinton couldn't be as sure about OpenAI, though he has never worked at the company. When asked whether the company's CEO, Sam Altman, has a "good moral compass" earlier in the episode, he said, "We'll see." He added that he doesn't know Altman personally, so he didn't want to comment further.

OpenAI has faced criticism in recent months for approaching safety differently than in the past. In a recent blog post, the company said it would only change its safety requirements after making sure it wouldn't "meaningfully increase the overall risk of severe harm." Its focus areas for safety now include cybersecurity, chemical threats, and AI's power to improve independently.

Altman defended OpenAI's approach to safety in an interview at TED2025 in April, saying that the company's preparedness framework outlines "where we think the most important danger moments are." Altman also acknowledged in the interview that OpenAI has loosened some restrictions on its model's behavior based on user feedback about censorship.

The earlier competition between OpenAI and Google to release initial chatbots was fierce, and the AI talent race is only heating up. Documents reviewed by Business Insider reveal that Google relied on ChatGPT in 2023 — during its attempts to catch up to ChatGPT.

Representatives for Google and OpenAI did not respond to BI's request for comment.

• An AI-generated ad for Kalshi, where you can bet on real-world events, aired during an NBA Finals game.
• PJ Accetturo, a self-described AI filmmaker, described his process for creating the ad.
• Here's how he used Google's Gemini chatbot and Veo 3 video generator to make the "most unhinged" ad.

A farmer floating in a pool of eggs. An alien chugging beer. An older man, draped in an American flag, screaming, "Indiana gonna win baby." The chaotic scenes are all part of a new AI-generated ad from sports betting marketplace Kalshi, which aired Wednesday during Game 3 of the NBA Finals.

"The world's gone mad, trade it," the commercial's tagline read, following the 30-second collection of surreal scenes.

In a recent thread on X, the ad's director explained how he made the clip for just $2,000.

"Kalshi hired me to make the most unhinged NBA Finals commercial possible," PJ Accetturo, a self-described AI filmmaker, wrote on Wednesday. "Network TV actually approved this GTA-style madness."

Accetturo said he made the ad using Veo 3, Google's latest AI video generator. A Kalshi spokesperson confirmed to BI that the company hired Accetturo to make the ad and that it was generated entirely using Veo 3.

"Kalshi asked me to create a spot about people betting on various markets, including the NBA Finals," Accetturo wrote on X. "I said the best Veo 3 content is crazy people doing crazy things while showcasing your brand. They love GTA VI. I grew up in Florida. This idea wrote itself."

He said that he started by writing a rough script, turned to Gemini to generate a shot list and prompts, pasted it into Veo 3, and made the finishing touches in editing software.

To write the script, he said he asked Kalshi's team for pieces of dialogue they wanted to include, then thought up "10 wild characters in unhinged situations to say them." Accetturo said that he got help from Gemini and ChatGPT for coming up with ideas and working them into a script.

A screenshot he posted of this stage of his process showed dialogue like "Indiana gonna win baby" and "I'm all in on OKC" alongside characters like "rizzed out grandpa headed to the club" and "old lady in front of pickup truck that says 'fresh manatee' in a cooler behind her."

Accetturo said he then asked Gemini to turn every shot description into a Veo 3 prompt.

"I always tell it to return 5 prompts at a time—any more than that and the quality starts to slip," he wrote on X. "Each prompt should fully describe the scene as if Veo 3 has no context of the shot before or after it. Re-describe the setting, the character, and the tone every time to maintain consistency."

Accetturo said it took 300 to 400 generations to get 15 usable clips.

"We were not specifically looking for an AI video at first, but after getting quotes from production companies that were in the six or seven figure range with timelines that didn't fit our needs, we decided to experiment, and that's when we made the decision to go with AI and hire PJ," the Kalshi spokesperson told BI. "Given the success of this first ad, we are absolutely planning on doing more with AI."

The spokesperson said the video went from idea to live ad in three days, cost roughly $2,000 to make, and is on track to finish with 20 million impressions across mediums.

Accetturo told BI that he was "paid very well for the project" and now makes a "lot more as an AI director" than he did for live action contracts, which often involved weeks of work before and after the shoot compared to the few days the Kalshi ad required.

"The client got an insane ad for a great rate on a blistering timeline, and I got paid really well, while working in my underwear," he said.

It seems like hardly a day goes by anymore without a new version of Google's Gemini AI landing, and sure enough, Google is rolling out a major update to its most powerful 2.5 Pro model. This release is aimed at fixing some problems that cropped up in an earlier Gemini Pro update, and the word is, this version will become a stable release that comes to the Gemini app for everyone to use.

The previous Gemini 2.5 Pro release, known as the I/O Edition, or simply 05-06, was focused on coding upgrades. Google claims the new version is even better at generating code, with a new high score of 82.2 percent in the Aider Polyglot test. That beats the best from OpenAI, Anthropic, and DeepSeek by a comfortable margin.

While the general-purpose Gemini 2.5 Flash has left preview, the Pro version is lagging behind. In fact, the last several updates have attracted some valid criticism of 2.5 Pro's performance outside of coding tasks since the big 03-25 update. Google's Logan Kilpatrick says the team has taken that feedback to heart and that the new model "closes [the] gap on 03-25 regressions." For example, users will supposedly see more creativity with better formatting of responses.

Read full article

Comments

One of the “godfathers” of artificial intelligence has attacked a multibillion-dollar race to develop the cutting-edge technology, saying the latest models are displaying dangerous characteristics such as lying to users.

Yoshua Bengio, a Canadian academic whose work has informed techniques used by top AI groups such as OpenAI and Google, said: “There’s unfortunately a very competitive race between the leading labs, which pushes them towards focusing on capability to make the AI more and more intelligent, but not necessarily put enough emphasis and investment on research on safety.”

The Turing Award winner issued his warning in an interview with the Financial Times, while launching a new non-profit called LawZero. He said the group would focus on building safer systems, vowing to “insulate our research from those commercial pressures.”

Read full article

Comments

You may not use Gemini or other AI products, but many people do, and their ranks are growing. During day three of Google's antitrust remedies trial, the company presented a slide showing that Gemini reached 350 million monthly active users as of March 2025. That's a massive increase from last year, showing that Google is beginning to gain traction among competing chatbots, but Google's estimation of ChatGPT's traffic shows it still has a long climb ahead of it.

The slide was presented during the testimony of Sissie Hsiao, who until recently was leading Google's Gemini efforts. She was replaced earlier this month by Josh Woodward, who also runs Google Labs. The slide listed Gemini's 350 million monthly users, along with daily traffic of 35 million users.

These numbers represent a huge increase for Gemini, which languished in the tens of millions of monthly users late last year. Gemini's daily user count at the time was a mere 9 million, according to Google. Since then, Google has released its Gemini 2.0 and 2.5 models, both of which have shown demonstrable improvements over the previous iterations. It has also begun adding Gemini features to more parts of the Google ecosystem, even though some of those integrations can be more frustrating than useful.

Read full article

Comments

On Wednesday, OpenAI announced the release of two new models—o3 and o4-mini—that combine simulated reasoning capabilities with access to functions like web browsing and coding. These models mark the first time OpenAI's reasoning-focused models can use every ChatGPT tool simultaneously, including visual analysis and image generation.

OpenAI announced o3 in December, and until now, only less capable derivative models named "o3-mini" and "03-mini-high" have been available. However, the new models replace their predecessors—o1 and o3-mini.

OpenAI is rolling out access today for ChatGPT Plus, Pro, and Team users, with Enterprise and Edu customers gaining access next week. Free users can try o4-mini by selecting the "Think" option before submitting queries. OpenAI CEO Sam Altman tweeted that "we expect to release o3-pro to the pro tier in a few weeks."

Read full article

Comments

Google made waves with the release of Gemini 2.5 last month, rocketing to the top of the AI leaderboard after previously struggling to keep up with the likes of OpenAI. That first experimental model was just the beginning. Google is deploying its improved AI in more places across its ecosystem, from the developer-centric Vertex AI to the consumer Gemini app.

Gemini models have been dropping so quickly, it can be hard to grasp Google's intended lineup. Things are becoming clearer now that the company is beginning to move its products to the new branch. At the Google Cloud Next conference, it has announced initial availability of Gemini 2.5 Flash. This model is based on the same code as Gemini 2.5 Pro, but it's faster and cheaper to run.

You won't see Gemini 2.5 Flash in the Gemini app just yet—it's starting out in the Vertex AI development platform. The experimental wide release of Pro helped Google gather data and see how people interacted with the new model, and that has helped inform the development of 2.5 Flash.

Read full article

Comments