What sort of industry connections could buy influence over a scientist’s research results? Research grants as small as US$5,000? Money to support outreach that bolsters the industry’s image? Equity in a spin-off company founded by the scientist? Defining what constitutes a conflict of interest — much less regulating it — continues to vex funding agencies, journals and institutions. Last month, for instance, Nature revealed that an activist organization had filed freedom-of-information requests to see the e-mails of researchers who work on genetically modified crops. Among other findings, their haul revealed that plant scientist Kevin Folta at the University of Florida in Gainesville had accepted a no-strings-attached$25,000 grant from the agriculture giant Monsanto to fund travel for a communications training programme about biotechnology, as well as other travel reimbursements from the biotech industry. In his defence, Folta argued that he had complied with the disclosure rules set out by the University of Florida. There is no evidence of any wrongdoing or that his research was compromised.

Solar physicist Willie Soon, a climate-change sceptic at the Harvard-Smithsonian Center for Astrophysics in Massachusetts, also seems to have been operating within institutional policy when advocacy groups revealed in February that he had accepted more than \$1 million from the energy industry, among other funders. (However, his failure to disclose those relationships might have violated the policies of some journals in which he published.)

In trying to navigate such complexities, the US National Institutes of Health (NIH) has been ahead of the curve — presumably because of long-standing concerns about physicians’ industry relationships and the high stakes for protecting patients. Its parent agency, the Department of Health and Human Services (HHS), was the first to establish conflict-of-interest disclosure rules in 1995 and is still beyond many of its counterparts in maintaining unified regulations that include yearly reports to the government. By contrast, as one example, the US National Science Foundation’s grants policy suggests that institutions look to scientific societies for ideas on how to manage a conflict of interest, and to report back to the foundation only if institutions cannot handle it themselves.

But even the HHS rules were not enough to guarantee full transparency. In 2009, a congressional report and subsequent media coverage found that some NIH-funded researchers had quietly accepted millions of dollars from industry. Again, the blame kept shifting: the universities said that the researchers had not reported the conflicts, the NIH received only bare-bones reports from institutions, and the researchers said that they did not know they were breaking any rules.

The reforms may not be perfect, but others should take note.

The HHS updated its policies in 2011, but pleased no one. The government underestimated the time and money that institutions would spend implementing new rules. And some aspects of the reforms have proved to be window dressing: a Nature investigation this week reveals that these reforms have uncovered few conflicts of interest that would have escaped the original regulations (see page 300).

The reforms may not be perfect, but they address real issues and others should take note. They make it clear that institutions are accountable, that they must educate their researchers on financial disclosure and that they should evaluate whether an industry relationship is problematic. The reforms also enlist a second pair of eyes by requiring institutions to report details of the conflict and its management to the NIH. Perhaps most importantly, the reforms remove the excuse of plausible deniability by clearly stating the kinds of financial relationship that could be considered conflicts.

One thing has become clear: conflicts are slippery to define, so it is important for as many funders, institutions and journals to make as many demands as necessary. Had Kevin Folta been funded by the NIH, the HHS guidelines would have required him to report the Monsanto money. And if Willie Soon had had an NIH grant, his institution would have designed a ‘management plan’ that could have required his industry relationships to be stated in publications and lectures.

The HHS rules could backfire. Institutions do not want the publicity and work that accompany an identified conflict. Because they hold the power to decide whether a relationship presents a conflict, they could theoretically give their researchers a pass. Nature’s investigation suggests that institutions use vastly different standards to evaluate such relationships, meaning that the rule is unevenly applied. And the current system makes it difficult for the public to access the conflict reports.

Still, the HHS should be commended for at least attempting to address the problem, even if it was forced into doing so. Other funders and institutions could do worse than to learn from its successes and mistakes if they define and strengthen their own policies.