The AI was wrong. She paid the fine.

Christina Weaver of Kodiak, Alaska, pleaded no contest and paid a $150 fine for hunting snipe out of season. The Alaska Beacon reported it on September 28. The opening date she hunted on came from Google.

She had typed 2 words into her phone: snipe season. The AI Overview, the box that sits above the actual search results, said September 1. On September 5 she took 3 birds. Then she learned duck season did not open until October 8, and that did not line up. She read the state regulations. For her, the answer was off by more than a month. She called dispatch and turned herself in. Trooper Alex Wick ran the same search and got the same wrong answer.

The state's public safety spokesman said she did everything right once she caught it. He said it is the first time he thinks they have seen AI cited, and the department does not track it. Google did not respond when the Beacon asked. So this is only the first one anybody wrote down.

That was the thread all week. The answer came from a machine. The bill went to the person who acted on it.

A phone showing an empty search bar resting on a backlit keyboard

Photo: Zulfugar Karimov via Pexels

What changed

A camera logged her car correctly. She spent 13 days in jail anyway. Lindsey Isaacs, 23, was charged with 8 felonies after a Flock camera recorded her black Durango 2 minutes before a fatal Florida crash, about 3 miles away. A 911 caller had described a maroon one. Prosecutors dropped every charge in May. She told her story to a Senate subcommittee. The CEOs of Flock, Axon, Motorola, and Verkada were invited. None came. She is suing 2 troopers, not Flock, and Flock disputes that its technology caused her arrest.

Kenjiro Tsuda lost his case against TikTok. The court still said a voice is protected. An anonymous account posted 188 videos in a voice the anime actor says is an AI copy of his. On September 30 the Tokyo District Court dismissed his request to remove them, because they were already deleted. Judge Aya Takahashi wrote that using a performer's voice without permission can violate publicity rights. It is being reported as the first Japanese ruling to say that about a voice.

A performer in headphones recording at a studio microphone

Photo: Tima Miroshnichenko via Pexels

A message asked for the wire. A phone call confirmed it. Both were fake. Sources told Reuters that in February Paolo Molesini, then chairman of Fideuram, Intesa Sanpaolo's private bank, got a WhatsApp that appeared to come from the group CEO, then a call in what the sources say was an AI copy of a senior lawyer's voice. He told his finance team to pay. €95 million went out. About €53 million came back. Roughly €36 million is still gone. Molesini is not under investigation. The bank declined to comment.

Hands holding a phone with a bank fraud alert on the lock screen

Photo: RDNE Stock project via Pexels

A gambler says an AI picked him for at least 70 promotions in one month. Daniel Vest of West Virginia filed a proposed class action against DraftKings on September 30. It leans on a New York Times investigation in which former employees described a model that scored casino players on how much they were expected to lose after a bonus. DraftKings denies it, says it does not use AI to target customers based on losses, and will fight the suit. Massachusetts regulators say they will look at how sportsbooks use AI.

Outside the channel. On September 29, CNBC reported, President Trump said he and tech executives signed a "morally binding" AI document after a White House lunch, and that he is "seeing tremendous self-policing." House Speaker Mike Johnson called it a statement of principles that is voluntary for the industry. The next day California went the other way. Governor Newsom signed SB 947, which bars employers from relying solely on automated systems to fire or discipline a worker.

What it means for your business

Look at who was left holding it this week. Weaver paid the fine, not Google. A bank is still missing roughly €36 million because a phone call sounded right. OpenAI says its own agents posted 53 users' photos online and it cannot find the people to tell them. When AI gets it wrong, the one who acted on it answers for it. So set up your shop that way on purpose.

Five moves for this week.

Click past the summary for anything with a date, a dollar figure, or a rule. Permit deadlines, tax dates, building codes, license renewals. The AI box at the top of a search is not the source. Give your team one line: if it goes on a form or a calendar, someone opens the agency's own page and reads it there.

Rewrite your wire approval so the confirmation call goes out, never in. A callback only counts if you place it yourself, to a number you already had on file. A text from the boss plus a call confirming it is 2 channels a scammer can control. Put the rule in writing and hand it to everyone who can approve a payment by phone.

Put a named person between AI output and any firing or discipline. California's new law requires human review and telling the worker when an automated system was used. Connecticut, as of October 1, says an AI decision tool is not a defense to a discrimination claim. Wherever you are, write down who reviewed the output and what they checked.

Get it in writing who answers when a vendor's AI acts on its own. Before you sign or renew, ask one question: if your tool takes an action and it causes harm, who is responsible under this contract? If the answer is not in the contract, the answer is you.

Put one name next to every AI tool your team logs into. List what people really use, who approved it, and what goes into it. That name is the person who answers for what the tool can see. If a tool has no name next to it, nobody is checking it.

None of this needs new software. It needs an hour on Monday and a name on every line.

What your team should and should not put into a chatbot, in plain terms: Is ChatGPT Safe to Use at Work? A Straight Answer for Business Owners.

Two people reviewing and signing printed documents at a table

Photo: Mikhail Nilov via Pexels

What to ignore

"AI did the layoffs." Employers cited AI in 116,175 announced job cuts in 2026 through August, according to the outplacement firm Challenger, Gray and Christmas. It is the leading reason companies have given this year. New York added a question to its mass layoff filing asking whether automation played a part. In the first year, more than 160 companies filed. Not one said yes.

It changes nothing Monday. A press release is not a filing. Plan your hiring on your own numbers, not on what another company told its investors. And if a budget deck in your office says a tool saves 6 positions, know that Connecticut now asks that question by statute.