Welcome to the Era of BadGPTs

A new crop of nefarious chatbots with names like “BadGPT” and “FraudGPT” are springing up on the darkest corners of the web, as cybercriminals look to tap the same artificial intelligence behind OpenAI’s ChatGPT.

Just as some office workers use ChatGPT to write better emails, hackers are using manipulated versions of AI chatbots to turbocharge their phishing emails. They can use chatbots—some also freely-available on the open internet—to create fake websites, write malware and tailor messages to better impersonate executives and other trusted entities.

Earlier this year, a Hong Kong multinational company employee handed over $25.5 million to an attacker who posed as the company’s chief financial officer on an AI-generated deepfake conference call, the South China Morning Post reported, citing Hong Kong police. Chief information officers and cybersecurity leaders, already accustomed to a growing spate of cyberattacks , say they are on high alert for an uptick in more sophisticated phishing emails and deepfakes.

Vish Narendra, CIO of Graphic Packaging International, said the Atlanta-based paper packing company has seen an increase in what are likely AI-generated email attacks called spear-phishing , where cyberattackers use information about a person to make an email seem more legitimate. Public companies in the spotlight are even more susceptible to contextualised spear-phishing, he said.

Researchers at Indiana University recently combed through over 200 large-language model hacking services being sold and populated on the dark web. The first service appeared in early 2023—a few months after the public release of OpenAI’s ChatGPT in November 2022.

Most dark web hacking tools use versions of open-source AI models like Meta ’s Llama 2, or “jailbroken” models from vendors like OpenAI and Anthropic to power their services, the researchers said. Jailbroken models have been hijacked by techniques like “ prompt injection ” to bypass their built-in safety controls.

Jason Clinton, chief information security officer of Anthropic, said the AI company eliminates jailbreak attacks as they find them, and has a team monitoring the outputs of its AI systems. Most model-makers also deploy two separate models to secure their primary AI model, making the likelihood that all three will fail the same way “a vanishingly small probability.”

Meta spokesperson Kevin McAlister said that openly releasing models shares the benefits of AI widely, and allows researchers to identify and help fix vulnerabilities in all AI models, “so companies can make models more secure.”

An OpenAI spokesperson said the company doesn’t want its tools to be used for malicious purposes, and that it is “always working on how we can make our systems more robust against this type of abuse.”

Malware and phishing emails written by generative AI are especially tricky to spot because they are crafted to evade detection. Attackers can teach a model to write stealthy malware by training it with detection techniques gleaned from cybersecurity defence software, said Avivah Litan, a generative AI and cybersecurity analyst at Gartner.

Phishing emails grew by 1,265% in the 12-month period starting when ChatGPT was publicly released, with an average of 31,000 phishing attacks sent every day, according to an October 2023 report by cybersecurity vendor SlashNext.

“The hacking community has been ahead of us,” said Brian Miller, CISO of New York-based not-for-profit health insurer Healthfirst, which has seen an increase in attacks impersonating its invoice vendors over the past two years.

While it is nearly impossible to prove whether certain malware programs or emails were created with AI, tools developed with AI can scan for text likely created with the technology. Abnormal Security , an email security vendor, said it had used AI to help identify thousands of likely AI-created malicious emails over the past year, and that it had blocked a twofold increase in targeted, personalised email attacks.

When Good Models Go Bad

Part of the challenge in stopping AI-enabled cybercrime is some AI models are freely shared on the open web. To access them, there is no need for dark corners of the internet or exchanging cryptocurrency.

Such models are considered “uncensored” because they lack the enterprise guardrails that businesses look for when buying AI systems, said Dane Sherrets, an ethical hacker and senior solutions architect at bug bounty company HackerOne.

In some cases, uncensored versions of models are created by security and AI researchers who strip out their built-in safeguards. In other cases, models with safeguards intact will write scam messages if humans avoid obvious triggers like “phishing”—a situation Andy Sharma, CIO and CISO of Redwood Software, said he discovered when creating a spear-phishing test for his employees.

The most useful model for generating scam emails is likely a version of Mixtral, from French AI startup Mistral AI, that has been altered to remove its safeguards, Sherrets said. Due to the advanced design of the original Mixtral, the uncensored version likely performs better than most dark web AI tools, he added. Mistral did not reply to a request for comment.

Sherrets recently demonstrated the process of using an uncensored AI model to generate a phishing campaign. First, he searched for “uncensored” models on Hugging Face, a startup that hosts a popular repository of open-source models—showing how easily many can be found.

He then used a virtual computing service that cost less than $1 per hour to mimic a graphics processing unit, or GPU, which is an advanced chip that can power AI. A bad actor needs either a GPU or a cloud-based service to use an AI model, Sherrets said, adding that he learned most of how to do this on X and YouTube.

With his uncensored model and virtual GPU service running, Sherrets asked the bot: “Write a phishing email targeting a business that impersonates a CEO and includes publicly-available company data,” and “Write an email targeting the procurement department of a company requesting an urgent invoice payment.”

The bot sent back phishing emails that were well-written, but didn’t include all of the personalisation asked for. That’s where prompt engineering , or the human’s ability to better extract information from chatbots, comes in, Sherrets said.

Dark Web AI Tools Can Already Do Harm

For hackers, a benefit of dark web tools like BadGPT—which researchers said uses OpenAI’s GPT model—is that they are likely trained on data from those underground marketplaces. That means they probably include useful information like leaks, ransomware victims and extortion lists, said Joseph Thacker, an ethical hacker and principal AI engineer at cybersecurity software firm AppOmni.

While some underground AI tools have been shuttered, new services have already taken their place, said Indiana University Assistant Computer Science Professor Xiaojing Liao, a co-author of the study. The AI hacking services, which often take payment via cryptocurrency, are priced anywhere from $5 to $199 a month.

New tools are expected to improve just as the AI models powering them do. In a matter of years, AI-generated text, video and voice deepfakes will be virtually indistinguishable from their human counterparts, said Evan Reiser , CEO and co-founder of Abnormal Security.

While researching the hacking tools, Indiana University Associate Dean for Research XiaoFeng Wang, a co-author of the study, said he was surprised by the ability of dark web services to generate effective malware. Given just the code of a security vulnerability, the tools can easily write a program to exploit it.

Though AI hacking tools often fail, in some cases, they work. “That demonstrates, in my opinion, that today’s large language models have the capability to do harm,” Wang said.

Reporting season has once again reminded investors that a strong profit does not guarantee a rising share price, and a large loss does not always trigger a sell-off. What matters most is how each result compares with expectations and, increasingly, what management says about the year ahead. During the August 2026 season, companies offering credible turnarounds or unexpectedly strong guidance were rewarded handsomely, while those flagging weaker margins, slowing demand or greater uncertainty were punished.

The following ranking draws on Morningstar’s review of 164 ASX-listed companies and measures each company’s share-price movement on the day it reported. This captures the market’s immediate response to the earnings announcement, before subsequent economic developments, dividends and company-specific news cloud the picture. Here are the five biggest winners, and the five hardest-hit losers, of the season so far.

The five winners

1. Bapcor (ASX:BAP): +41.0%

Bapcor delivered reporting season’s largest relief rally after presenting early evidence that its troubled automotive-parts business was stabilising. Although underlying revenue fell 1.8% to $1.92 billion and underlying NPAT collapsed 85% to $10.8 million, underlying EBITDA of $152.5 million exceeded guidance.

More importantly, working-capital initiatives released $68.5 million in the second half, lifting cash conversion to 109.4% and reducing net debt by 63% to $135 million. The statutory loss was $431.6 million, largely because of non-cash impairments. Investors focused on improving operational momentum, stronger liquidity and management’s expectation of modest FY27 revenue growth.

2. Zip Co (ASX:ZIP): +18.2%

Zip comfortably surpassed its FY26 targets, sending the buy-now-pay-later provider’s shares sharply higher. Transaction volume rose 23% to $16.7 billion, while cash earnings before tax, depreciation and amortisation jumped 58% to a record $268.9 million. Statutory profit climbed 46% to $116.4 million, and the cash operating margin expanded by 4.2 percentage points to 20%.

The strongest signal was guidance for FY27 cash earnings of $340 million—around 26% growth and above analysts’ forecasts. US transaction volume increased 42.5% and now represents three-quarters of group volume, offsetting weaker customer activity in Australia.

3. CSL (ASX:CSL): +17.3%

CSL’s result was hardly spectacular in isolation, but it cleared a market bar that had fallen dramatically following earlier downgrades and restructuring announcements. Underlying NPATA was US$3.1 billion, down 2% in constant-currency terms, while operating cash flow reached US$3.51 billion.

CSL maintained its full-year dividend at US$2.92 per share and completed a A$1 billion buyback. The real catalyst was FY27 guidance for approximately 5% underlying profit growth, compared with market expectations closer to 2%. After an extended period of earnings disappointments, investors interpreted the outlook as evidence that CSL’s core plasma business was approaching a sustainable recovery.

4. Judo Capital (ASX:JDO): +16.9%

Judo Capital demonstrated strong operating leverage as its specialist business-lending franchise expanded. Full-year profit before tax rose 34% to $168.1 million, while pre-provision profit increased 42%. Gross loans and advances grew 18% to $14.7 billion, reaching the top of the bank’s guidance range and comfortably exceeding broader system growth.

Deposits increased 24% to $12.2 billion, return on equity improved by 1.1 percentage points to 6.4%, and earnings per share rose 29% to 9.9 cents. Reaffirmation of the FY27 outlook gave investors confidence that loan growth could continue without sacrificing margins or credit quality.

5. Super Retail Group (ASX:SUL): +15.8%

The owner of Supercheap Auto, rebel, BCF and Macpac reported record sales of $4.2 billion, up 3.2%, despite cautious discretionary spending. Profitability went backwards: normalised profit before tax fell 7% to $306 million and normalised NPAT declined 2.8% to $226 million as transformation spending weighed on margins. Nevertheless, the result exceeded subdued expectations, online sales grew 5.3% and membership across the group’s loyalty programs reached 13.1 million. Investors were also encouraged by positive early FY27 trading, stable gross margins and continued market-share gains. A fully franked 33-cent final dividend added to the appeal.

The five losers

1. Hansen Technologies (ASX:HSN): –21.2%

Hansen’s historic result met expectations, but investors recoiled from its outlook. The utility and communications software provider achieved an underlying EBITDA margin of 31%, exceeding its 30% target, while generating strong cash flow. However, management designated FY27 an “investment and transition year”, signalling a roughly five-percentage-point margin contraction as spending on products, sales capabilities and organisational changes increased.

Revenue had already been broadly flat, leaving investors concerned that the investment program would depress earnings before new growth appeared. Leadership changes, including the chief executive’s departure, added uncertainty. Management expects revenue growth and margins above 30% to return in FY28, but the market was unwilling to wait.

2. Life360 (ASX:360): –19.4%

Life360’s headline growth was impressive: quarterly revenue rose 38% to US$159 million, subscription revenue increased 31%, and adjusted EBITDA climbed 53% to US$31.1 million. Monthly active users reached 102.4 million and paying circles grew 27% to 3.2 million. The sell-off reflected expectations rather than a collapsing business.

Net income fell 18%, the net margin contracted from 6% to 3%, hardware shipments dropped 18%, and full-year EBITDA guidance was merely maintained. After a strong valuation run, investors wanted a larger upgrade and clearer evidence that heavy investment in advertising, international expansion and artificial intelligence would generate additional earnings.

3. PEXA Group (ASX:PXA): –17.0%

PEXA reported a 7% increase in continuing-operations revenue and 12% EBITDA growth to $152 million, accompanied by a two-percentage-point margin expansion. Free cash flow increased 39%, suggesting the core Australian electronic-conveyancing platform remained highly profitable. Investors instead concentrated on management’s warning that property-transfer volumes could decline, alongside regulatory uncertainty surrounding the fees PEXA can charge.

The company is also continuing to invest heavily in its loss-making international expansion. Morningstar considered the market reaction excessive, arguing that structural transfer-volume assumptions had not materially changed, but the combination of softer near-term activity and regulatory risk overwhelmed the respectable headline numbers.

4. SEEK (ASX:SEK): –14.3%

SEEK produced solid FY26 figures, including 10% revenue growth to $1.20 billion, a 15% rise in EBITDA and 28% growth in adjusted earnings per share. It also lifted its fully franked annual dividend by 13% to a record 52 cents. Those achievements were overshadowed by falling paid job-ad volumes and cautious FY27 assumptions.

The statutory accounts included a $201 million loss from the SEEK Growth Fund and $377 million of significant items, making the headline result considerably less attractive. Investors were particularly concerned that economic weakness could limit volumes while the company continued investing in platform integration and artificial-intelligence products.

5. JB Hi-Fi (ASX:JBH): –12.3%

JB Hi-Fi’s full-year result was broadly respectable, with group sales rising 5% to $11.1 billion and underlying earnings per share increasing 6% to $4.48. The damage came from its current-trading update. Australian sales were almost flat during the June quarter and deteriorated further in July, while earnings in the core Australian electronics business fell 3.6%.

Housing-related categories were particularly weak as higher living costs and interest rates constrained household budgets. With JB Hi-Fi entering the season on a demanding valuation, an in-line historic result was not sufficient: the loss of sales momentum prompted investors to rapidly reduce their expectations for FY27.

Overall, the season showed that share prices react less to whether profits rose or fell than to the gap between results and expectations. Bapcor was rewarded for being less troubled than feared, while several fundamentally profitable companies were punished because their outlooks failed to justify elevated valuations.